{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 6.177, "avg": 6.18511724, "p5": 5.993, "p95": 6.363, "min": 5.916, "max": 6.513, "stddev": 0.116, "init": 12.699, "first_inference": 16.051, "warm_up_avg": 6.09782934}, "memory_mb": {"init_footprint": 29.89, "overall_footprint": 38.43}, "runs": 162, "warmup_runs": 82, "throughput_mb_s": 92.8387833, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-6fa889ae", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 0.956, "avg": 0.966870546, "p5": 0.944, "p95": 1.006, "min": 0.934, "max": 1.724, "stddev": 0.055, "init": 524.062, "first_inference": 5.662, "warm_up_avg": 0.979839504}, "memory_mb": {"init_footprint": 94.32, "overall_footprint": 94.32}, "runs": 1035, "warmup_runs": 511, "throughput_mb_s": 593.894165, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-89479d22", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/ios/iphone17pro/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "ios", "device_id": "iphone17pro", "device": "iPhone 17 Pro", "os": "iOS 27.0 (24A437)", "accelerator": "cpu", "delegate": "XNNPACK", "nodes_delegated": null, "nodes_total": null, "partitions": null, "latency_ms": {"median": 2.304, "avg": 2.35844946, "p5": 2.164, "p95": 2.771, "min": 2.057, "max": 6.919, "stddev": 0.302, "init": 15.921, "first_inference": 4.751, "warm_up_avg": 2.47495079}, "memory_mb": {"init_footprint": 15.62, "overall_footprint": 25.72}, "runs": 425, "warmup_runs": 203, "throughput_mb_s": 243.472992, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), app in benchmark/ios", "runner": "app", "session": "2026-09-17-iphone17pro-default", "job": "cpu-iphone17pro", "status": "measured", "date": "2026-09-17"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/ios/iphone17pro/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "ios", "device_id": "iphone17pro", "device": "iPhone 17 Pro", "os": "iOS 27.0 (24A437)", "accelerator": "gpu", "delegate": "Metal", "nodes_delegated": null, "nodes_total": null, "partitions": null, "latency_ms": {"median": 0.82, "avg": 0.839021802, "p5": 0.759, "p95": 0.892, "min": 0.734, "max": 2.355, "stddev": 0.119, "init": 503.228, "first_inference": 6.405, "warm_up_avg": 0.892016053}, "memory_mb": {"init_footprint": 36.27, "overall_footprint": 60.05}, "runs": 1192, "warmup_runs": 561, "throughput_mb_s": 684.390747, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), app in benchmark/ios", "runner": "app", "session": "2026-09-17-iphone17pro-default", "job": "gpu-iphone17pro", "status": "measured", "date": "2026-09-17"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 2.053, "avg": 2.06332779, "p5": 2.034, "p95": 2.124, "min": 2.027, "max": 2.269, "stddev": 0.032, "init": 27.434, "first_inference": 3.204, "warm_up_avg": 2.06812811}, "memory_mb": {"init_footprint": 15.03, "overall_footprint": 23.0}, "runs": 485, "warmup_runs": 242, "throughput_mb_s": 278.297394, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001757", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 0.45, "avg": 0.457295388, "p5": 0.438, "p95": 0.497, "min": 0.417, "max": 0.777, "stddev": 0.035, "init": 42.93, "first_inference": 4.3, "warm_up_avg": 0.482410789}, "memory_mb": {"init_footprint": 35.47, "overall_footprint": 217.73}, "runs": 2187, "warmup_runs": 1037, "throughput_mb_s": 1255.68457, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001757", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_int8_channelwise.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_int8_channelwise.tflite", "task": "image-classification", "model_size_mb": 4.00552, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 1.648, "avg": 1.65681791, "p5": 1.639, "p95": 1.698, "min": 1.636, "max": 1.857, "stddev": 0.022, "init": 9.027, "first_inference": 2.047, "warm_up_avg": 1.68797982}, "memory_mb": {"init_footprint": 6.55, "overall_footprint": 8.52}, "runs": 604, "warmup_runs": 297, "throughput_mb_s": 86.6448212, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001800", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_int8_channelwise.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_int8_channelwise.tflite", "task": "image-classification", "model_size_mb": 4.00552, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 0.561, "avg": 0.570055306, "p5": 0.549, "p95": 0.621, "min": 0.537, "max": 0.911, "stddev": 0.04, "init": 302.655, "first_inference": 4.463, "warm_up_avg": 0.589182556}, "memory_mb": {"init_footprint": 26.77, "overall_footprint": 209.59}, "runs": 1754, "warmup_runs": 849, "throughput_mb_s": 251.825897, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001800", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_dynamic_wi8_afp32.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_dynamic_wi8_afp32.tflite", "task": "image-classification", "model_size_mb": 3.88779211, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 53, "nodes_total": 70, "partitions": 35, "latency_ms": {"median": 4.972, "avg": 5.01014, "p5": 4.906, "p95": 5.256, "min": 4.883, "max": 5.829, "stddev": 0.12, "init": 7.842, "first_inference": 5.832, "warm_up_avg": 4.97232676}, "memory_mb": {"init_footprint": 5.14, "overall_footprint": 20.62}, "runs": 200, "warmup_runs": 101, "throughput_mb_s": 114.61132, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001803", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_dynamic_wi8_afp32.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_dynamic_wi8_afp32.tflite", "task": "image-classification", "model_size_mb": 3.88779211, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 0.449, "avg": 0.456448197, "p5": 0.441, "p95": 0.486, "min": 0.431, "max": 0.745, "stddev": 0.035, "init": 39.095, "first_inference": 4.021, "warm_up_avg": 0.486774325}, "memory_mb": {"init_footprint": 25.69, "overall_footprint": 206.97}, "runs": 2191, "warmup_runs": 1028, "throughput_mb_s": 1258.01514, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001803", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-small:mobilenet_v3_small.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/MobileNet-v3-small", "file": "mobilenet_v3_small.tflite", "task": "image-classification", "model_size_mb": 10.1815205, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 122, "nodes_total": 122, "partitions": 1, "latency_ms": {"median": 0.898, "avg": 0.900690377, "p5": 0.89, "p95": 0.918, "min": 0.86, "max": 0.981, "stddev": 0.01, "init": 13.771, "first_inference": 1.166, "warm_up_avg": 0.901059449}, "memory_mb": {"init_footprint": 11.73, "overall_footprint": 15.14}, "runs": 1111, "warmup_runs": 555, "throughput_mb_s": 637.531799, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001807", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-small:mobilenet_v3_small.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/MobileNet-v3-small", "file": "mobilenet_v3_small.tflite", "task": "image-classification", "model_size_mb": 10.1815205, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 122, "nodes_total": 122, "partitions": 1, "latency_ms": {"median": 0.457, "avg": 0.465722859, "p5": 0.447, "p95": 0.517, "min": 0.432, "max": 0.77, "stddev": 0.035, "init": 336.777, "first_inference": 4.575, "warm_up_avg": 0.471304417}, "memory_mb": {"init_footprint": 29.61, "overall_footprint": 209.05}, "runs": 2147, "warmup_runs": 1061, "throughput_mb_s": 1232.96228, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001807", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-large:mobilenet_v3_large.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/MobileNet-v3-large", "file": "mobilenet_v3_large.tflite", "task": "image-classification", "model_size_mb": 21.9215832, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 134, "nodes_total": 134, "partitions": 1, "latency_ms": {"median": 2.102, "avg": 2.10956, "p5": 2.077, "p95": 2.166, "min": 2.057, "max": 2.343, "stddev": 0.032, "init": 27.634, "first_inference": 2.629, "warm_up_avg": 2.11674261}, "memory_mb": {"init_footprint": 23.56, "overall_footprint": 29.73}, "runs": 475, "warmup_runs": 237, "throughput_mb_s": 272.198364, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001810", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-large:mobilenet_v3_large.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/MobileNet-v3-large", "file": "mobilenet_v3_large.tflite", "task": "image-classification", "model_size_mb": 21.9215832, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 134, "nodes_total": 134, "partitions": 1, "latency_ms": {"median": 0.527, "avg": 0.534927249, "p5": 0.514, "p95": 0.575, "min": 0.499, "max": 0.881, "stddev": 0.038, "init": 253.273, "first_inference": 4.328, "warm_up_avg": 0.547768891}, "memory_mb": {"init_footprint": 52.55, "overall_footprint": 233.78}, "runs": 1870, "warmup_runs": 913, "throughput_mb_s": 1073.45203, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001810", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_b0:efficientnet_b0.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/efficientnet_b0", "file": "efficientnet_b0.tflite", "task": "image-classification", "model_size_mb": 21.1264477, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 244, "nodes_total": 244, "partitions": 1, "latency_ms": {"median": 6.796, "avg": 6.8759656, "p5": 6.691, "p95": 7.401, "min": 6.647, "max": 8.571, "stddev": 0.294, "init": 29.027, "first_inference": 7.785, "warm_up_avg": 6.80541897}, "memory_mb": {"init_footprint": 23.73, "overall_footprint": 35.44}, "runs": 146, "warmup_runs": 74, "throughput_mb_s": 83.510994, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001813", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_b0:efficientnet_b0.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/efficientnet_b0", "file": "efficientnet_b0.tflite", "task": "image-classification", "model_size_mb": 21.1264477, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 244, "nodes_total": 244, "partitions": 1, "latency_ms": {"median": 0.741, "avg": 0.751811266, "p5": 0.718, "p95": 0.809, "min": 0.707, "max": 1.101, "stddev": 0.046, "init": 269.722, "first_inference": 4.705, "warm_up_avg": 0.762047172}, "memory_mb": {"init_footprint": 53.19, "overall_footprint": 235.39}, "runs": 1330, "warmup_runs": 657, "throughput_mb_s": 763.780457, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001813", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_v2_s:efficientnet_v2_s.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/efficientnet_v2_s", "file": "efficientnet_v2_s.tflite", "task": "image-classification", "model_size_mb": 85.6423492, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 508, "nodes_total": 508, "partitions": 1, "latency_ms": {"median": 37.039, "avg": 37.3238, "p5": 36.791, "p95": 39.993, "min": 36.741, "max": 40.219, "stddev": 0.831, "init": 118.143, "first_inference": 39.523, "warm_up_avg": 38.1122131}, "memory_mb": {"init_footprint": 88.56, "overall_footprint": 119.38}, "runs": 50, "warmup_runs": 14, "throughput_mb_s": 45.2124367, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001817", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_v2_s:efficientnet_v2_s.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/efficientnet_v2_s", "file": "efficientnet_v2_s.tflite", "task": "image-classification", "model_size_mb": 85.6423492, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 508, "nodes_total": 508, "partitions": 1, "latency_ms": {"median": 3.003, "avg": 3.02482176, "p5": 2.965, "p95": 3.235, "min": 2.957, "max": 3.369, "stddev": 0.076, "init": 454.797, "first_inference": 8.562, "warm_up_avg": 3.07522082}, "memory_mb": {"init_footprint": 191.66, "overall_footprint": 361.14}, "runs": 331, "warmup_runs": 163, "throughput_mb_s": 557.884094, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001817", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet18:resnet18.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/resnet18", "file": "resnet18.tflite", "task": "image-classification", "model_size_mb": 46.7505608, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 37, "nodes_total": 37, "partitions": 1, "latency_ms": {"median": 3.743, "avg": 3.7600565, "p5": 3.674, "p95": 3.901, "min": 3.653, "max": 4.085, "stddev": 0.07, "init": 75.016, "first_inference": 4.772, "warm_up_avg": 3.86936927}, "memory_mb": {"init_footprint": 45.94, "overall_footprint": 58.98}, "runs": 266, "warmup_runs": 130, "throughput_mb_s": 152.715469, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001821", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet18:resnet18.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/resnet18", "file": "resnet18.tflite", "task": "image-classification", "model_size_mb": 46.7505608, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 37, "nodes_total": 37, "partitions": 1, "latency_ms": {"median": 0.947, "avg": 0.955837607, "p5": 0.927, "p95": 1.025, "min": 0.905, "max": 1.248, "stddev": 0.04, "init": 278.468, "first_inference": 4.45, "warm_up_avg": 0.965071321}, "memory_mb": {"init_footprint": 113.92, "overall_footprint": 294.78}, "runs": 1047, "warmup_runs": 519, "throughput_mb_s": 600.749268, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001821", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet50:resnet50.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/resnet50", "file": "resnet50.tflite", "task": "image-classification", "model_size_mb": 102.147186, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 78, "nodes_total": 78, "partitions": 1, "latency_ms": {"median": 8.474, "avg": 8.49077129, "p5": 8.338, "p95": 8.65, "min": 8.27, "max": 9.607, "stddev": 0.146, "init": 155.264, "first_inference": 9.385, "warm_up_avg": 8.47028351}, "memory_mb": {"init_footprint": 99.97, "overall_footprint": 116.06}, "runs": 118, "warmup_runs": 60, "throughput_mb_s": 67.6285706, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001825", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet50:resnet50.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/resnet50", "file": "resnet50.tflite", "task": "image-classification", "model_size_mb": 102.147186, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 78, "nodes_total": 78, "partitions": 1, "latency_ms": {"median": 1.818, "avg": 1.83534861, "p5": 1.798, "p95": 1.979, "min": 1.781, "max": 2.119, "stddev": 0.057, "init": 402.362, "first_inference": 7.206, "warm_up_avg": 1.86891794}, "memory_mb": {"init_footprint": 222.02, "overall_footprint": 404.02}, "runs": 545, "warmup_runs": 268, "throughput_mb_s": 312.866302, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001825", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/squeezenet1_1:squeezenet1_1.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/squeezenet1_1", "file": "squeezenet1_1.tflite", "task": "image-classification", "model_size_mb": 4.95489597, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 40, "nodes_total": 40, "partitions": 1, "latency_ms": {"median": 1.068, "avg": 1.07797956, "p5": 1.059, "p95": 1.119, "min": 1.051, "max": 1.46, "stddev": 0.03, "init": 8.842, "first_inference": 1.512, "warm_up_avg": 1.07305789}, "memory_mb": {"init_footprint": 5.86, "overall_footprint": 12.06}, "runs": 928, "warmup_runs": 466, "throughput_mb_s": 532.680542, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001828", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/squeezenet1_1:squeezenet1_1.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/squeezenet1_1", "file": "squeezenet1_1.tflite", "task": "image-classification", "model_size_mb": 4.95489597, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 40, "nodes_total": 40, "partitions": 1, "latency_ms": {"median": 0.383, "avg": 0.389380068, "p5": 0.376, "p95": 0.407, "min": 0.367, "max": 0.675, "stddev": 0.031, "init": 147.611, "first_inference": 3.371, "warm_up_avg": 0.410321563}, "memory_mb": {"init_footprint": 18.11, "overall_footprint": 182.3}, "runs": 2568, "warmup_runs": 1219, "throughput_mb_s": 1474.7, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001828", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/shufflenet_v2_x0_5:shufflenet_v2_x0_5.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/shufflenet_v2_x0_5", "file": "shufflenet_v2_x0_5.tflite", "task": "image-classification", "model_size_mb": 5.64969206, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 189, "nodes_total": 190, "partitions": 3, "latency_ms": {"median": 0.616, "avg": 0.619882286, "p5": 0.609, "p95": 0.643, "min": 0.603, "max": 0.687, "stddev": 0.011, "init": 8.905, "first_inference": 1.067, "warm_up_avg": 0.624704123}, "memory_mb": {"init_footprint": 7.52, "overall_footprint": 12.67}, "runs": 1614, "warmup_runs": 801, "throughput_mb_s": 926.335144, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001831", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/shufflenet_v2_x0_5:shufflenet_v2_x0_5.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/shufflenet_v2_x0_5", "file": "shufflenet_v2_x0_5.tflite", "task": "image-classification", "model_size_mb": 5.64969206, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 10, "nodes_total": 190, "partitions": 3, "latency_ms": {"median": 0.754, "avg": 0.759284735, "p5": 0.728, "p95": 0.802, "min": 0.714, "max": 1.277, "stddev": 0.039, "init": 98.285, "first_inference": 4.479, "warm_up_avg": 0.767236173}, "memory_mb": {"init_footprint": 14.86, "overall_footprint": 183.12}, "runs": 1317, "warmup_runs": 652, "throughput_mb_s": 756.262756, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001831", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/googlenet:googlenet.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/googlenet", "file": "googlenet.tflite", "task": "image-classification", "model_size_mb": 26.5059204, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 93, "nodes_total": 105, "partitions": 25, "latency_ms": {"median": 3.583, "avg": 3.6016655, "p5": 3.536, "p95": 3.72, "min": 3.525, "max": 3.942, "stddev": 0.064, "init": 43.95, "first_inference": 4.678, "warm_up_avg": 3.59360719}, "memory_mb": {"init_footprint": 27.7, "overall_footprint": 43.19}, "runs": 278, "warmup_runs": 140, "throughput_mb_s": 159.431458, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001834", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/googlenet:googlenet.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/googlenet", "file": "googlenet.tflite", "task": "image-classification", "model_size_mb": 26.5059204, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 105, "nodes_total": 105, "partitions": 1, "latency_ms": {"median": 1.041, "avg": 1.05800629, "p5": 1.028, "p95": 1.154, "min": 1.013, "max": 1.427, "stddev": 0.057, "init": 208.067, "first_inference": 5.335, "warm_up_avg": 1.09680045}, "memory_mb": {"init_footprint": 73.81, "overall_footprint": 255.66}, "runs": 946, "warmup_runs": 456, "throughput_mb_s": 542.736572, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001834", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/inception_v3:inception_v3.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/inception_v3", "file": "inception_v3.tflite", "task": "image-classification", "model_size_mb": 95.5243, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 139, "nodes_total": 141, "partitions": 5, "latency_ms": {"median": 11.104, "avg": 11.1431446, "p5": 10.944, "p95": 11.493, "min": 10.897, "max": 11.654, "stddev": 0.175, "init": 145.513, "first_inference": 12.291, "warm_up_avg": 11.0939569}, "memory_mb": {"init_footprint": 94.73, "overall_footprint": 112.34}, "runs": 90, "warmup_runs": 46, "throughput_mb_s": 91.8154907, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001838", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/inception_v3:inception_v3.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/inception_v3", "file": "inception_v3.tflite", "task": "image-classification", "model_size_mb": 95.5243, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 141, "nodes_total": 141, "partitions": 1, "latency_ms": {"median": 2.883, "avg": 2.9135406, "p5": 2.862, "p95": 3.107, "min": 2.842, "max": 3.184, "stddev": 0.071, "init": 330.405, "first_inference": 8.158, "warm_up_avg": 2.96503544}, "memory_mb": {"init_footprint": 211.91, "overall_footprint": 381.05}, "runs": 344, "warmup_runs": 169, "throughput_mb_s": 351.158051, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001838", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/convnext_tiny:convnext_tiny.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/convnext_tiny", "file": "convnext_tiny.tflite", "task": "image-classification", "model_size_mb": 114.425186, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 160, "nodes_total": 183, "partitions": 47, "latency_ms": {"median": 52.242, "avg": 52.3868217, "p5": 51.682, "p95": 53.019, "min": 51.544, "max": 56.156, "stddev": 0.771, "init": 145.235, "first_inference": 53.453, "warm_up_avg": 52.4494}, "memory_mb": {"init_footprint": 113.48, "overall_footprint": 134.67}, "runs": 50, "warmup_runs": 10, "throughput_mb_s": 10.9611301, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001841", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/convnext_tiny:convnext_tiny.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/convnext_tiny", "file": "convnext_tiny.tflite", "task": "image-classification", "model_size_mb": 114.425186, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 183, "nodes_total": 183, "partitions": 1, "latency_ms": {"median": 1.946, "avg": 1.9622333, "p5": 1.925, "p95": 2.149, "min": 1.902, "max": 2.338, "stddev": 0.058, "init": 359.022, "first_inference": 7.065, "warm_up_avg": 2.00151205}, "memory_mb": {"init_footprint": 242.78, "overall_footprint": 424.53}, "runs": 510, "warmup_runs": 250, "throughput_mb_s": 292.635315, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001841", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/vit_base_patch16_224.augreg_in1k:vit_base_patch16_224_fp32.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/vit_base_patch16_224.augreg_in1k", "file": "vit_base_patch16_224_fp32.tflite", "task": "image-classification", "model_size_mb": 346.216125, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 363, "nodes_total": 400, "partitions": 27, "latency_ms": {"median": 44.906, "avg": 45.1213608, "p5": 44.576, "p95": 46.076, "min": 44.51, "max": 48.814, "stddev": 0.76, "init": 413.357, "first_inference": 46.398, "warm_up_avg": 45.2010841}, "memory_mb": {"init_footprint": 333.5, "overall_footprint": 348.84}, "runs": 50, "warmup_runs": 12, "throughput_mb_s": 12.7260962, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001847", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/vit_base_patch16_224.augreg_in1k:vit_base_patch16_224_fp32.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/vit_base_patch16_224.augreg_in1k", "file": "vit_base_patch16_224_fp32.tflite", "task": "image-classification", "model_size_mb": 346.216125, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 28, "nodes_total": 400, "partitions": 3, "latency_ms": {"median": 44.086, "avg": 44.2713814, "p5": 43.606, "p95": 44.893, "min": 43.564, "max": 48.611, "stddev": 0.818, "init": 237.693, "first_inference": 46.733, "warm_up_avg": 44.1826668}, "memory_mb": {"init_footprint": 355.61, "overall_footprint": 525.58}, "runs": 50, "warmup_runs": 12, "throughput_mb_s": 12.9704285, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001847", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/lraspp_mobilenet_v3_large:lraspp_mobilenet_v3_large.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/lraspp_mobilenet_v3_large", "file": "lraspp_mobilenet_v3_large.tflite", "task": "image-segmentation", "model_size_mb": 12.8802719, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 156, "nodes_total": 156, "partitions": 1, "latency_ms": {"median": 17.559, "avg": 17.6770344, "p5": 17.33, "p95": 18.421, "min": 17.22, "max": 20.5, "stddev": 0.464, "init": 17.665, "first_inference": 20.925, "warm_up_avg": 17.635725}, "memory_mb": {"init_footprint": 14.98, "overall_footprint": 82.66}, "runs": 57, "warmup_runs": 29, "throughput_mb_s": 175.056641, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001853", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/lraspp_mobilenet_v3_large:lraspp_mobilenet_v3_large.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/lraspp_mobilenet_v3_large", "file": "lraspp_mobilenet_v3_large.tflite", "task": "image-segmentation", "model_size_mb": 12.8802719, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 156, "nodes_total": 156, "partitions": 1, "latency_ms": {"median": 1.308, "avg": 1.33612013, "p5": 1.294, "p95": 1.443, "min": 1.267, "max": 10.094, "stddev": 0.326, "init": 356.677, "first_inference": 5.815, "warm_up_avg": 1.34994066}, "memory_mb": {"init_footprint": 36.2, "overall_footprint": 252.05}, "runs": 749, "warmup_runs": 371, "throughput_mb_s": 2316.021, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001853", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/deeplabv3_resnet50:deeplabv3_resnet50.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/deeplabv3_resnet50", "file": "deeplabv3_resnet50.tflite", "task": "image-segmentation", "model_size_mb": 158.472351, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 86, "nodes_total": 87, "partitions": 3, "latency_ms": {"median": 243.441, "avg": 244.079437, "p5": 242.068, "p95": 250.213, "min": 241.615, "max": 254.613, "stddev": 2.446, "init": 222.846, "first_inference": 254.194, "warm_up_avg": 246.095337}, "memory_mb": {"init_footprint": 153.89, "overall_footprint": 360.86}, "runs": 50, "warmup_runs": 3, "throughput_mb_s": 12.6781769, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001856", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/deeplabv3_resnet50:deeplabv3_resnet50.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/deeplabv3_resnet50", "file": "deeplabv3_resnet50.tflite", "task": "image-segmentation", "model_size_mb": 158.472351, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 87, "nodes_total": 87, "partitions": 1, "latency_ms": {"median": 24.637, "avg": 24.6005, "p5": 24.429, "p95": 24.757, "min": 24.426, "max": 24.805, "stddev": 0.105, "init": 690.542, "first_inference": 35.42, "warm_up_avg": 25.1226501}, "memory_mb": {"init_footprint": 362.66, "overall_footprint": 590.16}, "runs": 50, "warmup_runs": 20, "throughput_mb_s": 125.789413, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001856", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-tiny:whisper_tiny_30s_f32.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/whisper-tiny", "file": "whisper_tiny_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 150.979187, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 73, "nodes_total": 82, "partitions": 18, "latency_ms": {"median": 59.96, "avg": 60.17062, "p5": 59.45, "p95": 60.743, "min": 59.223, "max": 65.745, "stddev": 1.065, "init": 172.395, "first_inference": 68.402, "warm_up_avg": 61.318779}, "memory_mb": {"init_footprint": 147.67, "overall_footprint": 222.06}, "runs": 50, "warmup_runs": 9, "throughput_mb_s": 15.2155209, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001912", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-tiny:whisper_tiny_30s_f32.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/whisper-tiny", "file": "whisper_tiny_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 150.979187, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 82, "nodes_total": 82, "partitions": 1, "latency_ms": {"median": 4.342, "avg": 4.37196493, "p5": 4.316, "p95": 4.584, "min": 4.309, "max": 4.629, "stddev": 0.08, "init": 766.111, "first_inference": 11.275, "warm_up_avg": 4.43523026}, "memory_mb": {"init_footprint": 576.55, "overall_footprint": 772.16}, "runs": 229, "warmup_runs": 113, "throughput_mb_s": 209.408661, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001912", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-base:whisper_base_30s_f32.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/whisper-base", "file": "whisper_base_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 290.082642, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 343, "nodes_total": 343, "partitions": 1, "latency_ms": {"median": 126.434, "avg": 126.782242, "p5": 125.908, "p95": 129.524, "min": 125.71, "max": 130.014, "stddev": 0.963, "init": 313.816, "first_inference": 134.816, "warm_up_avg": 129.055496}, "memory_mb": {"init_footprint": 280.84, "overall_footprint": 370.06}, "runs": 50, "warmup_runs": 4, "throughput_mb_s": 7.22125864, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001918", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-base:whisper_base_30s_f32.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/whisper-base", "file": "whisper_base_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 290.082642, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 343, "nodes_total": 343, "partitions": 1, "latency_ms": {"median": 12.886, "avg": 12.9026413, "p5": 12.739, "p95": 13.137, "min": 12.735, "max": 13.208, "stddev": 0.131, "init": 1289.00806, "first_inference": 21.274, "warm_up_avg": 13.1024876}, "memory_mb": {"init_footprint": 837.24, "overall_footprint": 1075.3}, "runs": 78, "warmup_runs": 39, "throughput_mb_s": 70.9565811, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001918", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/moonshine-tiny:moonshine_tiny_5s_f32.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/moonshine-tiny", "file": "moonshine_tiny_5s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 109.373138, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 321, "nodes_total": 358, "partitions": 38, "latency_ms": {"median": 10.994, "avg": 11.0352306, "p5": 10.827, "p95": 11.439, "min": 10.729, "max": 11.655, "stddev": 0.174, "init": 140.427, "first_inference": 11.614, "warm_up_avg": 11.2827559}, "memory_mb": {"init_footprint": 113.98, "overall_footprint": 123.8}, "runs": 91, "warmup_runs": 45, "throughput_mb_s": 27.6546803, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001928", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/moonshine-tiny:moonshine_tiny_5s_f32.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/moonshine-tiny", "file": "moonshine_tiny_5s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 109.373138, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 358, "nodes_total": 358, "partitions": 1, "latency_ms": {"median": 2.082, "avg": 2.11464262, "p5": 2.066, "p95": 2.35, "min": 2.055, "max": 2.448, "stddev": 0.08, "init": 921.748, "first_inference": 7.188, "warm_up_avg": 2.17274022}, "memory_mb": {"init_footprint": 329.25, "overall_footprint": 495.11}, "runs": 473, "warmup_runs": 231, "throughput_mb_s": 144.315536, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001928", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/Gecko-110m-en:Gecko_256_f32.tflite@2.2.0/macos/mac-studio-m4-max/cpu", "model": "litert-community/Gecko-110m-en", "file": "Gecko_256_f32.tflite", "task": "question-answering", "model_size_mb": 443.197906, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 614, "nodes_total": 633, "partitions": 2, "latency_ms": {"median": 43.955, "avg": 44.1300201, "p5": 43.64, "p95": 45.173, "min": 43.622, "max": 48.072, "stddev": 0.711, "init": 409.44, "first_inference": 48.257, "warm_up_avg": 44.5229149}, "memory_mb": {"init_footprint": 333.03, "overall_footprint": 363.09}, "runs": 50, "warmup_runs": 12, "throughput_mb_s": 0.0221292097, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001932", "job": "cpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/Gecko-110m-en:Gecko_256_f32.tflite@2.2.0/macos/mac-studio-m4-max/gpu", "model": "litert-community/Gecko-110m-en", "file": "Gecko_256_f32.tflite", "task": "question-answering", "model_size_mb": 443.197906, "platform": "macos", "device_id": "mac-studio-m4-max", "device": "Mac Studio (M4 Max)", "os": "macOS 27.0", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 589, "nodes_total": 633, "partitions": 2, "latency_ms": {"median": 7.257, "avg": 7.298594, "p5": 7.191, "p95": 7.523, "min": 7.172, "max": 7.58, "stddev": 0.102, "init": 1328.67102, "first_inference": 16.67, "warm_up_avg": 7.4268384}, "memory_mb": {"init_footprint": 496.8, "overall_footprint": 694.94}, "runs": 138, "warmup_runs": 68, "throughput_mb_s": 0.133801445, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/macos_arm64/benchmark_model", "runner": "local", "session": "local-20260918-001932", "job": "gpu-mac-studio-m4-max", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_int8_channelwise.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_int8_channelwise.tflite", "task": "image-classification", "model_size_mb": 4.00552, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 2.26, "avg": 2.22577786, "p5": 2.037, "p95": 2.332, "min": 1.999, "max": 2.392, "stddev": 0.093, "init": 4.762, "first_inference": 3.185, "warm_up_avg": 2.24048209}, "memory_mb": {"init_footprint": 11.0, "overall_footprint": 14.38}, "runs": 450, "warmup_runs": 224, "throughput_mb_s": 64.4964142, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-376e611c", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_int8_channelwise.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_int8_channelwise.tflite", "task": "image-classification", "model_size_mb": 4.00552, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 1.283, "avg": 1.4585743, "p5": 1.164, "p95": 1.924, "min": 1.106, "max": 2.256, "stddev": 0.285, "init": 538.413, "first_inference": 6.558, "warm_up_avg": 1.67894626}, "memory_mb": {"init_footprint": 74.4, "overall_footprint": 74.4}, "runs": 686, "warmup_runs": 298, "throughput_mb_s": 98.4212341, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-ccc5f129", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_dynamic_wi8_afp32.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_dynamic_wi8_afp32.tflite", "task": "image-classification", "model_size_mb": 3.88779211, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 53, "nodes_total": 70, "partitions": 35, "latency_ms": {"median": 4.446, "avg": 4.43854427, "p5": 4.281, "p95": 4.586, "min": 4.192, "max": 4.746, "stddev": 0.099, "init": 5.972, "first_inference": 6.176, "warm_up_avg": 4.60366964}, "memory_mb": {"init_footprint": 10.62, "overall_footprint": 26.69}, "runs": 226, "warmup_runs": 109, "throughput_mb_s": 129.370972, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-225dd834", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_dynamic_wi8_afp32.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_dynamic_wi8_afp32.tflite", "task": "image-classification", "model_size_mb": 3.88779211, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 1.121, "avg": 1.17759478, "p5": 0.987, "p95": 1.483, "min": 0.942, "max": 1.911, "stddev": 0.15, "init": 505.253, "first_inference": 6.048, "warm_up_avg": 1.15217972}, "memory_mb": {"init_footprint": 73.34, "overall_footprint": 73.34}, "runs": 849, "warmup_runs": 434, "throughput_mb_s": 487.619965, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-2a518f62", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-small:mobilenet_v3_small.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/MobileNet-v3-small", "file": "mobilenet_v3_small.tflite", "task": "image-classification", "model_size_mb": 10.1815205, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 122, "nodes_total": 122, "partitions": 1, "latency_ms": {"median": 1.869, "avg": 1.84879303, "p5": 1.699, "p95": 1.931, "min": 1.612, "max": 1.974, "stddev": 0.072, "init": 4.858, "first_inference": 2.659, "warm_up_avg": 1.81844}, "memory_mb": {"init_footprint": 23.11, "overall_footprint": 27.11}, "runs": 541, "warmup_runs": 275, "throughput_mb_s": 310.591156, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-b64996ac", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-small:mobilenet_v3_small.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/MobileNet-v3-small", "file": "mobilenet_v3_small.tflite", "task": "image-classification", "model_size_mb": 10.1815205, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 122, "nodes_total": 122, "partitions": 1, "latency_ms": {"median": 0.724, "avg": 0.73097, "p5": 0.617, "p95": 1.037, "min": 0.598, "max": 1.66, "stddev": 0.124, "init": 573.972, "first_inference": 6.16, "warm_up_avg": 0.779046714}, "memory_mb": {"init_footprint": 86.23, "overall_footprint": 86.23}, "runs": 1368, "warmup_runs": 642, "throughput_mb_s": 785.55719, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-2ec2499a", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-large:mobilenet_v3_large.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/MobileNet-v3-large", "file": "mobilenet_v3_large.tflite", "task": "image-classification", "model_size_mb": 21.9215832, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 134, "nodes_total": 134, "partitions": 1, "latency_ms": {"median": 5.933, "avg": 5.95094061, "p5": 5.738, "p95": 6.239, "min": 5.646, "max": 6.534, "stddev": 0.156, "init": 8.011, "first_inference": 6.659, "warm_up_avg": 5.81467438}, "memory_mb": {"init_footprint": 45.57, "overall_footprint": 52.5}, "runs": 169, "warmup_runs": 86, "throughput_mb_s": 96.4920959, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-fe79ade6", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v3-large:mobilenet_v3_large.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/MobileNet-v3-large", "file": "mobilenet_v3_large.tflite", "task": "image-classification", "model_size_mb": 21.9215832, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 134, "nodes_total": 134, "partitions": 1, "latency_ms": {"median": 1.155, "avg": 1.22497427, "p5": 1.035, "p95": 1.544, "min": 0.984, "max": 3.749, "stddev": 0.218, "init": 806.696, "first_inference": 2.123, "warm_up_avg": 1.2254926}, "memory_mb": {"init_footprint": 116.42, "overall_footprint": 116.42}, "runs": 816, "warmup_runs": 408, "throughput_mb_s": 468.759857, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-a2db8511", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_b0:efficientnet_b0.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/efficientnet_b0", "file": "efficientnet_b0.tflite", "task": "image-classification", "model_size_mb": 21.1264477, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 244, "nodes_total": 244, "partitions": 1, "latency_ms": {"median": 13.535, "avg": 13.5384188, "p5": 13.29, "p95": 13.881, "min": 13.24, "max": 13.911, "stddev": 0.177, "init": 8.665, "first_inference": 14.866, "warm_up_avg": 13.3286581}, "memory_mb": {"init_footprint": 44.6, "overall_footprint": 57.13}, "runs": 74, "warmup_runs": 38, "throughput_mb_s": 42.4140167, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-34bec112", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_b0:efficientnet_b0.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/efficientnet_b0", "file": "efficientnet_b0.tflite", "task": "image-classification", "model_size_mb": 21.1264477, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 244, "nodes_total": 244, "partitions": 1, "latency_ms": {"median": 1.594, "avg": 1.65646684, "p5": 1.501, "p95": 2.083, "min": 1.473, "max": 2.856, "stddev": 0.197, "init": 629.73, "first_inference": 6.44, "warm_up_avg": 1.76723671}, "memory_mb": {"init_footprint": 112.95, "overall_footprint": 112.95}, "runs": 604, "warmup_runs": 283, "throughput_mb_s": 346.65271, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-d766b0ec", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_v2_s:efficientnet_v2_s.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/efficientnet_v2_s", "file": "efficientnet_v2_s.tflite", "task": "image-classification", "model_size_mb": 85.6423492, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 508, "nodes_total": 508, "partitions": 1, "latency_ms": {"median": 171.471, "avg": 171.998383, "p5": 163.833, "p95": 182.111, "min": 161.844, "max": 188.763, "stddev": 5.398, "init": 31.762, "first_inference": 160.521, "warm_up_avg": 159.66925}, "memory_mb": {"init_footprint": 170.01, "overall_footprint": 204.1}, "runs": 50, "warmup_runs": 4, "throughput_mb_s": 9.81113911, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-ea678b3b", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/efficientnet_v2_s:efficientnet_v2_s.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/efficientnet_v2_s", "file": "efficientnet_v2_s.tflite", "task": "image-classification", "model_size_mb": 85.6423492, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 508, "nodes_total": 508, "partitions": 1, "latency_ms": {"median": 7.841, "avg": 7.81324816, "p5": 7.458, "p95": 8.238, "min": 7.326, "max": 8.39, "stddev": 0.252, "init": 899.655, "first_inference": 12.458, "warm_up_avg": 7.87928104}, "memory_mb": {"init_footprint": 272.43, "overall_footprint": 272.43}, "runs": 129, "warmup_runs": 64, "throughput_mb_s": 215.979324, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-82a9cb8c", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet18:resnet18.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/resnet18", "file": "resnet18.tflite", "task": "image-classification", "model_size_mb": 46.7505608, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 37, "nodes_total": 37, "partitions": 1, "latency_ms": {"median": 31.92, "avg": 31.8797, "p5": 31.192, "p95": 32.48, "min": 30.884, "max": 32.556, "stddev": 0.424, "init": 28.487, "first_inference": 30.685, "warm_up_avg": 30.246}, "memory_mb": {"init_footprint": 92.07, "overall_footprint": 105.61}, "runs": 50, "warmup_runs": 17, "throughput_mb_s": 18.0120506, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-66fbbac3", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet18:resnet18.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/resnet18", "file": "resnet18.tflite", "task": "image-classification", "model_size_mb": 46.7505608, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 37, "nodes_total": 37, "partitions": 1, "latency_ms": {"median": 3.465, "avg": 3.60126972, "p5": 3.364, "p95": 4.077, "min": 3.244, "max": 7.499, "stddev": 0.369, "init": 517.329, "first_inference": 8.039, "warm_up_avg": 3.43895197}, "memory_mb": {"init_footprint": 170.03, "overall_footprint": 170.03}, "runs": 278, "warmup_runs": 146, "throughput_mb_s": 159.448975, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-4c5e12a8", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet50:resnet50.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/resnet50", "file": "resnet50.tflite", "task": "image-classification", "model_size_mb": 102.147186, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 78, "nodes_total": 78, "partitions": 1, "latency_ms": {"median": 72.325, "avg": 71.9471436, "p5": 69.9, "p95": 72.966, "min": 69.776, "max": 73.044, "stddev": 0.939, "init": 49.877, "first_inference": 70.475, "warm_up_avg": 68.84375}, "memory_mb": {"init_footprint": 198.32, "overall_footprint": 213.11}, "runs": 50, "warmup_runs": 8, "throughput_mb_s": 7.98112, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-21031579", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/resnet50:resnet50.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/resnet50", "file": "resnet50.tflite", "task": "image-classification", "model_size_mb": 102.147186, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 78, "nodes_total": 78, "partitions": 1, "latency_ms": {"median": 5.477, "avg": 5.51690674, "p5": 5.418, "p95": 5.641, "min": 5.407, "max": 7.91, "stddev": 0.2, "init": 706.964, "first_inference": 10.155, "warm_up_avg": 5.60102224}, "memory_mb": {"init_footprint": 302.99, "overall_footprint": 302.99}, "runs": 182, "warmup_runs": 90, "throughput_mb_s": 104.083466, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-3b55ad2f", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/squeezenet1_1:squeezenet1_1.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/squeezenet1_1", "file": "squeezenet1_1.tflite", "task": "image-classification", "model_size_mb": 4.95489597, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 40, "nodes_total": 40, "partitions": 1, "latency_ms": {"median": 5.922, "avg": 5.94131374, "p5": 5.752, "p95": 6.147, "min": 5.676, "max": 6.219, "stddev": 0.112, "init": 4.995, "first_inference": 7.759, "warm_up_avg": 6.18667889}, "memory_mb": {"init_footprint": 12.5, "overall_footprint": 19.25}, "runs": 169, "warmup_runs": 81, "throughput_mb_s": 96.6484528, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-96a1e95d", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/squeezenet1_1:squeezenet1_1.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/squeezenet1_1", "file": "squeezenet1_1.tflite", "task": "image-classification", "model_size_mb": 4.95489597, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 40, "nodes_total": 40, "partitions": 1, "latency_ms": {"median": 0.872, "avg": 0.935969114, "p5": 0.801, "p95": 1.288, "min": 0.789, "max": 1.546, "stddev": 0.149, "init": 531.341, "first_inference": 5.754, "warm_up_avg": 1.03518009}, "memory_mb": {"init_footprint": 72.18, "overall_footprint": 72.18}, "runs": 1069, "warmup_runs": 483, "throughput_mb_s": 613.501831, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-269bd967", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/shufflenet_v2_x0_5:shufflenet_v2_x0_5.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/shufflenet_v2_x0_5", "file": "shufflenet_v2_x0_5.tflite", "task": "image-classification", "model_size_mb": 5.64969206, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 189, "nodes_total": 190, "partitions": 3, "latency_ms": {"median": 1.345, "avg": 1.32451653, "p5": 1.179, "p95": 1.397, "min": 1.108, "max": 1.449, "stddev": 0.065, "init": 3.953, "first_inference": 2.493, "warm_up_avg": 1.30348432}, "memory_mb": {"init_footprint": 14.93, "overall_footprint": 20.8}, "runs": 755, "warmup_runs": 384, "throughput_mb_s": 433.530823, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-9346f70f", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/shufflenet_v2_x0_5:shufflenet_v2_x0_5.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/shufflenet_v2_x0_5", "file": "shufflenet_v2_x0_5.tflite", "task": "image-classification", "model_size_mb": 5.64969206, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 10, "nodes_total": 190, "partitions": 3, "latency_ms": {"median": 3.295, "avg": 3.42614388, "p5": 2.616, "p95": 4.33, "min": 2.007, "max": 7.262, "stddev": 0.715, "init": 259.898, "first_inference": 6.994, "warm_up_avg": 1.9934206}, "memory_mb": {"init_footprint": 62.21, "overall_footprint": 69.76}, "runs": 292, "warmup_runs": 252, "throughput_mb_s": 167.599136, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-dcee4098", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/googlenet:googlenet.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/googlenet", "file": "googlenet.tflite", "task": "image-classification", "model_size_mb": 26.5059204, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 93, "nodes_total": 105, "partitions": 25, "latency_ms": {"median": 25.86, "avg": 25.7867393, "p5": 25.351, "p95": 26.106, "min": 25.236, "max": 26.147, "stddev": 0.24, "init": 15.179, "first_inference": 26.373, "warm_up_avg": 24.975523}, "memory_mb": {"init_footprint": 53.94, "overall_footprint": 69.9}, "runs": 50, "warmup_runs": 21, "throughput_mb_s": 22.2679863, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-d0ac71a3", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/googlenet:googlenet.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/googlenet", "file": "googlenet.tflite", "task": "image-classification", "model_size_mb": 26.5059204, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 105, "nodes_total": 105, "partitions": 1, "latency_ms": {"median": 2.677, "avg": 2.69872499, "p5": 2.661, "p95": 2.811, "min": 2.653, "max": 2.968, "stddev": 0.05, "init": 530.408, "first_inference": 7.491, "warm_up_avg": 2.75560451}, "memory_mb": {"init_footprint": 123.54, "overall_footprint": 123.54}, "runs": 371, "warmup_runs": 182, "throughput_mb_s": 212.774078, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-c7b8aef0", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/inception_v3:inception_v3.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/inception_v3", "file": "inception_v3.tflite", "task": "image-classification", "model_size_mb": 95.5243, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 139, "nodes_total": 141, "partitions": 5, "latency_ms": {"median": 99.56, "avg": 99.1144638, "p5": 95.37, "p95": 101.155, "min": 94.754, "max": 101.317, "stddev": 1.823, "init": 52.811, "first_inference": 93.192, "warm_up_avg": 93.0371704}, "memory_mb": {"init_footprint": 186.4, "overall_footprint": 204.63}, "runs": 50, "warmup_runs": 6, "throughput_mb_s": 10.3225431, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-9a42fdc6", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/inception_v3:inception_v3.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/inception_v3", "file": "inception_v3.tflite", "task": "image-classification", "model_size_mb": 95.5243, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 141, "nodes_total": 141, "partitions": 1, "latency_ms": {"median": 7.937, "avg": 8.08087063, "p5": 7.824, "p95": 9.49, "min": 7.805, "max": 10.619, "stddev": 0.541, "init": 789.077, "first_inference": 12.659, "warm_up_avg": 8.05488873}, "memory_mb": {"init_footprint": 298.41, "overall_footprint": 298.41}, "runs": 124, "warmup_runs": 63, "throughput_mb_s": 126.609276, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-5e175284", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/convnext_tiny:convnext_tiny.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/convnext_tiny", "file": "convnext_tiny.tflite", "task": "image-classification", "model_size_mb": 114.425186, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 160, "nodes_total": 183, "partitions": 47, "latency_ms": {"median": 118.238, "avg": 118.22052, "p5": 113.53, "p95": 121.705, "min": 112.554, "max": 122.261, "stddev": 2.509, "init": 31.054, "first_inference": 117.929, "warm_up_avg": 112.545}, "memory_mb": {"init_footprint": 223.71, "overall_footprint": 243.14}, "runs": 50, "warmup_runs": 5, "throughput_mb_s": 4.85718346, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-89e79f3b", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/convnext_tiny:convnext_tiny.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/convnext_tiny", "file": "convnext_tiny.tflite", "task": "image-classification", "model_size_mb": 114.425186, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 183, "nodes_total": 183, "partitions": 1, "latency_ms": {"median": 4.574, "avg": 4.68854666, "p5": 4.498, "p95": 5.007, "min": 4.475, "max": 7.634, "stddev": 0.457, "init": 720.656, "first_inference": 9.458, "warm_up_avg": 4.66519451}, "memory_mb": {"init_footprint": 344.25, "overall_footprint": 344.25}, "runs": 214, "warmup_runs": 108, "throughput_mb_s": 122.472649, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-644787b6", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/vit_base_patch16_224.augreg_in1k:vit_base_patch16_224_fp32.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/vit_base_patch16_224.augreg_in1k", "file": "vit_base_patch16_224_fp32.tflite", "task": "image-classification", "model_size_mb": 346.216125, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 363, "nodes_total": 400, "partitions": 27, "latency_ms": {"median": 321.087, "avg": 319.393616, "p5": 304.139, "p95": 327.305, "min": 299.165, "max": 327.759, "stddev": 7.006, "init": 71.303, "first_inference": 288.018, "warm_up_avg": 291.735504}, "memory_mb": {"init_footprint": 663.37, "overall_footprint": 674.09}, "runs": 50, "warmup_runs": 2, "throughput_mb_s": 1.79784036, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-89abdf35", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/vit_base_patch16_224.augreg_in1k:vit_base_patch16_224_fp32.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/vit_base_patch16_224.augreg_in1k", "file": "vit_base_patch16_224_fp32.tflite", "task": "image-classification", "model_size_mb": 346.216125, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 28, "nodes_total": 400, "partitions": 3, "latency_ms": {"median": 313.444, "avg": 311.424835, "p5": 295.07, "p95": 322.451, "min": 290.758, "max": 323.081, "stddev": 8.137, "init": 475.982, "first_inference": 280.576, "warm_up_avg": 283.389496}, "memory_mb": {"init_footprint": 703.6, "overall_footprint": 712.61}, "runs": 50, "warmup_runs": 2, "throughput_mb_s": 1.84384394, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-577dd187", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/lraspp_mobilenet_v3_large:lraspp_mobilenet_v3_large.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/lraspp_mobilenet_v3_large", "file": "lraspp_mobilenet_v3_large.tflite", "task": "image-segmentation", "model_size_mb": 12.8802719, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 156, "nodes_total": 156, "partitions": 1, "latency_ms": {"median": 55.733, "avg": 55.6971588, "p5": 54.469, "p95": 56.751, "min": 54.204, "max": 57.308, "stddev": 0.685, "init": 7.119, "first_inference": 60.78, "warm_up_avg": 57.924221}, "memory_mb": {"init_footprint": 28.5, "overall_footprint": 94.84}, "runs": 50, "warmup_runs": 9, "throughput_mb_s": 55.5590706, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-64ad5caa", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/lraspp_mobilenet_v3_large:lraspp_mobilenet_v3_large.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/lraspp_mobilenet_v3_large", "file": "lraspp_mobilenet_v3_large.tflite", "task": "image-segmentation", "model_size_mb": 12.8802719, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 156, "nodes_total": 156, "partitions": 1, "latency_ms": {"median": 4.996, "avg": 5.03385448, "p5": 4.883, "p95": 5.365, "min": 4.869, "max": 5.756, "stddev": 0.163, "init": 944.349, "first_inference": 9.955, "warm_up_avg": 5.32556391}, "memory_mb": {"init_footprint": 96.16, "overall_footprint": 96.16}, "runs": 199, "warmup_runs": 94, "throughput_mb_s": 614.734192, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-5753b49b", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/deeplabv3_resnet50:deeplabv3_resnet50.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/deeplabv3_resnet50", "file": "deeplabv3_resnet50.tflite", "task": "image-segmentation", "model_size_mb": 158.472351, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 86, "nodes_total": 87, "partitions": 3, "latency_ms": {"median": 3216.11304, "avg": 3759.5769, "p5": 3019.27808, "p95": 5815.31885, "min": 2943.69189, "max": 5818.64111, "stddev": 1006.02698, "init": 86.516, "first_inference": 2824.08691, "warm_up_avg": 2824.08691}, "memory_mb": {"init_footprint": 305.79, "overall_footprint": 491.48}, "runs": 40, "warmup_runs": 1, "throughput_mb_s": 0.823093235, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-50878e0a", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/deeplabv3_resnet50:deeplabv3_resnet50.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/deeplabv3_resnet50", "file": "deeplabv3_resnet50.tflite", "task": "image-segmentation", "model_size_mb": 158.472351, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 87, "nodes_total": 87, "partitions": 1, "latency_ms": {"median": 77.075, "avg": 78.1111374, "p5": 73.432, "p95": 87.436, "min": 72.308, "max": 91.452, "stddev": 4.348, "init": 1076.57, "first_inference": 92.007, "warm_up_avg": 79.7314301}, "memory_mb": {"init_footprint": 459.7, "overall_footprint": 459.7}, "runs": 50, "warmup_runs": 7, "throughput_mb_s": 39.6164, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-06041579", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-tiny:whisper_tiny_30s_f32.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/whisper-tiny", "file": "whisper_tiny_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 150.979187, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 73, "nodes_total": 82, "partitions": 18, "latency_ms": {"median": 343.108, "avg": 340.363678, "p5": 324.245, "p95": 349.199, "min": 319.966, "max": 349.678, "stddev": 8.195, "init": 49.969, "first_inference": 318.572, "warm_up_avg": 318.432}, "memory_mb": {"init_footprint": 289.74, "overall_footprint": 364.05}, "runs": 50, "warmup_runs": 2, "throughput_mb_s": 2.68985033, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-046d6fea", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-tiny:whisper_tiny_30s_f32.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/whisper-tiny", "file": "whisper_tiny_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 150.979187, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 82, "nodes_total": 82, "partitions": 1, "latency_ms": {"median": 13.312, "avg": 13.468627, "p5": 13.184, "p95": 14.843, "min": 13.128, "max": 15.792, "stddev": 0.532, "init": 1071.76794, "first_inference": 18.178, "warm_up_avg": 13.8075943}, "memory_mb": {"init_footprint": 556.18, "overall_footprint": 556.18}, "runs": 75, "warmup_runs": 37, "throughput_mb_s": 67.9748077, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-2ed77391", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-base:whisper_base_30s_f32.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/whisper-base", "file": "whisper_base_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 290.082642, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 343, "nodes_total": 343, "partitions": 1, "latency_ms": {"median": 859.086, "avg": 851.27771, "p5": 796.323, "p95": 878.969, "min": 773.322, "max": 880.571, "stddev": 25.805, "init": 56.957, "first_inference": 757.502, "warm_up_avg": 757.502}, "memory_mb": {"init_footprint": 554.29, "overall_footprint": 646.29}, "runs": 50, "warmup_runs": 1, "throughput_mb_s": 1.07547438, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-86c9a07c", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/whisper-base:whisper_base_30s_f32.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/whisper-base", "file": "whisper_base_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 290.082642, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 343, "nodes_total": 343, "partitions": 1, "latency_ms": {"median": 51.657, "avg": 51.8944206, "p5": 50.816, "p95": 53.505, "min": 50.553, "max": 53.64, "stddev": 0.799, "init": 2260.31812, "first_inference": 55.513, "warm_up_avg": 53.398}, "memory_mb": {"init_footprint": 890.98, "overall_footprint": 890.98}, "runs": 50, "warmup_runs": 10, "throughput_mb_s": 17.6421146, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-5fceaf59", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/moonshine-tiny:moonshine_tiny_5s_f32.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/moonshine-tiny", "file": "moonshine_tiny_5s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 109.373138, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 321, "nodes_total": 358, "partitions": 38, "latency_ms": {"median": 42.989, "avg": 42.9212418, "p5": 42.113, "p95": 43.517, "min": 41.913, "max": 43.701, "stddev": 0.457, "init": 37.457, "first_inference": 42.577, "warm_up_avg": 41.5437698}, "memory_mb": {"init_footprint": 218.98, "overall_footprint": 228.59}, "runs": 50, "warmup_runs": 13, "throughput_mb_s": 7.11013412, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-8a8e2a10", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/moonshine-tiny:moonshine_tiny_5s_f32.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/moonshine-tiny", "file": "moonshine_tiny_5s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 109.373138, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 358, "nodes_total": 358, "partitions": 1, "latency_ms": {"median": 5.236, "avg": 5.34896803, "p5": 5.141, "p95": 5.709, "min": 5.119, "max": 5.993, "stddev": 0.209, "init": 1244.65405, "first_inference": 6.152, "warm_up_avg": 5.43490219}, "memory_mb": {"init_footprint": 371.23, "overall_footprint": 371.23}, "runs": 187, "warmup_runs": 92, "throughput_mb_s": 57.0532074, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-3126cbc1", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/Gecko-110m-en:Gecko_256_f32.tflite@2.2.0/android/pa3q-35/cpu", "model": "litert-community/Gecko-110m-en", "file": "Gecko_256_f32.tflite", "task": "question-answering", "model_size_mb": 443.197906, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 614, "nodes_total": 633, "partitions": 2, "latency_ms": {"median": 423.491, "avg": 420.818451, "p5": 399.361, "p95": 430.249, "min": 392.824, "max": 430.494, "stddev": 9.136, "init": 70.006, "first_inference": 379.549, "warm_up_avg": 383.83}, "memory_mb": {"init_footprint": 659.48, "overall_footprint": 691.55}, "runs": 50, "warmup_runs": 2, "throughput_mb_s": 0.00232062675, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-c6cb8e1c", "job": "cpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/Gecko-110m-en:Gecko_256_f32.tflite@2.2.0/android/pa3q-35/gpu", "model": "litert-community/Gecko-110m-en", "file": "Gecko_256_f32.tflite", "task": "question-answering", "model_size_mb": 443.197906, "platform": "android", "device_id": "pa3q-35", "device": "Galaxy S25 Ultra", "os": "Android 15 (API 35)", "accelerator": "gpu", "delegate": "LITERT_CL", "nodes_delegated": 589, "nodes_total": 633, "partitions": 2, "latency_ms": {"median": 45.405, "avg": 48.0378799, "p5": 39.632, "p95": 61.286, "min": 30.722, "max": 63.57, "stddev": 7.775, "init": 1009.453, "first_inference": 37.671, "warm_up_avg": 34.4832649}, "memory_mb": {"init_footprint": 880.19, "overall_footprint": 880.19}, "runs": 50, "warmup_runs": 15, "throughput_mb_s": 0.0203290097, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "2.2.0/android_arm64/benchmark_model", "runner": "ddp", "session": "session-8bbabce4", "job": "gpu-pa3q-35", "status": "measured", "date": "2026-09-18"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 2.158, "avg": 2.15919828, "p5": 2.04, "p95": 2.269, "min": 2.016, "max": 2.695, "stddev": 0.071, "init": 5.078, "first_inference": 3.875, "warm_up_avg": 2.2163229}, "memory_mb": {"init_footprint": 14.02, "overall_footprint": 21.84}, "runs": 464, "warmup_runs": 226, "throughput_mb_s": 265.940735, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-3a1e00a0-c0b3-41ed-81f0-0df6679a6435", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-09-30"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2.tflite", "task": "image-classification", "model_size_mb": 13.9759045, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 0.931, "avg": 0.933661044, "p5": 0.909, "p95": 0.967, "min": 0.894, "max": 1.055, "stddev": 0.018, "init": 382.679, "first_inference": 4.985, "warm_up_avg": 1.02998555}, "memory_mb": {"init_footprint": 28.05, "overall_footprint": 60.94}, "runs": 1071, "warmup_runs": 486, "throughput_mb_s": 615.018433, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-36f966b7-0eb6-4478-9540-8a729bdc0078", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-09-30"}
{"row_id": "litert-community/resnet50:resnet50.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/resnet50", "file": "resnet50.tflite", "task": "image-classification", "model_size_mb": 102.147186, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 78, "nodes_total": 78, "partitions": 1, "latency_ms": {"median": 9.257, "avg": 9.22757816, "p5": 8.976, "p95": 9.363, "min": 8.907, "max": 9.596, "stddev": 0.116, "init": 245.2, "first_inference": 17.841, "warm_up_avg": 9.45449066}, "memory_mb": {"init_footprint": 99.16, "overall_footprint": 121.16}, "runs": 109, "warmup_runs": 53, "throughput_mb_s": 62.2285461, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-c5ce1841-6ae4-4bba-93e9-27d30407d8f8", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/resnet50:resnet50.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/resnet50", "file": "resnet50.tflite", "task": "image-classification", "model_size_mb": 102.147186, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 78, "nodes_total": 78, "partitions": 1, "latency_ms": {"median": 5.585, "avg": 5.58502245, "p5": 5.535, "p95": 5.631, "min": 5.501, "max": 5.654, "stddev": 0.03, "init": 970.488, "first_inference": 17.013, "warm_up_avg": 5.64279795}, "memory_mb": {"init_footprint": 181.05, "overall_footprint": 215.02}, "runs": 180, "warmup_runs": 89, "throughput_mb_s": 102.814049, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-27727ed2-0d29-453a-9e8b-33559da0f106", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_int8_channelwise.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_int8_channelwise.tflite", "task": "image-classification", "model_size_mb": 4.00552, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 1.611, "avg": 1.59849834, "p5": 1.551, "p95": 1.626, "min": 1.549, "max": 1.744, "stddev": 0.028, "init": 12.067, "first_inference": 2.002, "warm_up_avg": 1.59830666}, "memory_mb": {"init_footprint": 5.41, "overall_footprint": 7.19}, "runs": 626, "warmup_runs": 313, "throughput_mb_s": 89.8059616, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-ad1cb639-8cee-443a-9291-8201e4eb12aa", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_int8_channelwise.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_int8_channelwise.tflite", "task": "image-classification", "model_size_mb": 4.00552, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 1.068, "avg": 1.11266291, "p5": 1.017, "p95": 1.309, "min": 0.988, "max": 1.358, "stddev": 0.1, "init": 460.564, "first_inference": 5.751, "warm_up_avg": 1.2116102}, "memory_mb": {"init_footprint": 19.16, "overall_footprint": 53.78}, "runs": 899, "warmup_runs": 413, "throughput_mb_s": 129.019028, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-b14c2eac-3b45-4e76-95dd-feb93ea26567", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_dynamic_wi8_afp32.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_dynamic_wi8_afp32.tflite", "task": "image-classification", "model_size_mb": 3.88779211, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 53, "nodes_total": 70, "partitions": 35, "latency_ms": {"median": 5.416, "avg": 5.40901089, "p5": 5.112, "p95": 5.635, "min": 5.1, "max": 6.181, "stddev": 0.142, "init": 14.797, "first_inference": 7.437, "warm_up_avg": 5.34789371}, "memory_mb": {"init_footprint": 4.06, "overall_footprint": 19.28}, "runs": 185, "warmup_runs": 94, "throughput_mb_s": 106.15966, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-76b6c38c-7664-4f4c-a10c-8d0dacbbf71f", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v2:mobilenet_v2_dynamic_wi8_afp32.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/MobileNet-v2", "file": "mobilenet_v2_dynamic_wi8_afp32.tflite", "task": "image-classification", "model_size_mb": 3.88779211, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 70, "nodes_total": 70, "partitions": 1, "latency_ms": {"median": 0.904, "avg": 0.903639555, "p5": 0.865, "p95": 0.941, "min": 0.839, "max": 1.013, "stddev": 0.021, "init": 393.895, "first_inference": 5.004, "warm_up_avg": 0.998135746}, "memory_mb": {"init_footprint": 18.77, "overall_footprint": 51.62}, "runs": 1107, "warmup_runs": 501, "throughput_mb_s": 635.451111, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-f834218d-b0ba-4ec0-aa09-faafff6418d1", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v3-small:mobilenet_v3_small.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/MobileNet-v3-small", "file": "mobilenet_v3_small.tflite", "task": "image-classification", "model_size_mb": 10.1815205, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 122, "nodes_total": 122, "partitions": 1, "latency_ms": {"median": 0.881, "avg": 0.881367385, "p5": 0.874, "p95": 0.89, "min": 0.872, "max": 0.918, "stddev": 0.005, "init": 4.13, "first_inference": 2.043, "warm_up_avg": 0.902054071}, "memory_mb": {"init_footprint": 10.81, "overall_footprint": 13.75}, "runs": 1135, "warmup_runs": 555, "throughput_mb_s": 651.509, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-6a74da1c-3e83-46f2-a50e-524476639e55", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v3-small:mobilenet_v3_small.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/MobileNet-v3-small", "file": "mobilenet_v3_small.tflite", "task": "image-classification", "model_size_mb": 10.1815205, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 122, "nodes_total": 122, "partitions": 1, "latency_ms": {"median": 0.897, "avg": 0.831317544, "p5": 0.62, "p95": 0.964, "min": 0.55, "max": 1.178, "stddev": 0.121, "init": 504.97, "first_inference": 3.687, "warm_up_avg": 0.736955822}, "memory_mb": {"init_footprint": 21.05, "overall_footprint": 51.83}, "runs": 1203, "warmup_runs": 679, "throughput_mb_s": 690.733337, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-c2b4b27d-b392-4c22-835c-01c3fd7657e8", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v3-large:mobilenet_v3_large.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/MobileNet-v3-large", "file": "mobilenet_v3_large.tflite", "task": "image-classification", "model_size_mb": 21.9215832, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 134, "nodes_total": 134, "partitions": 1, "latency_ms": {"median": 2.189, "avg": 2.29115343, "p5": 2.159, "p95": 2.75, "min": 2.153, "max": 4.817, "stddev": 0.267, "init": 6.401, "first_inference": 4.102, "warm_up_avg": 2.66088295}, "memory_mb": {"init_footprint": 22.48, "overall_footprint": 28.39}, "runs": 437, "warmup_runs": 188, "throughput_mb_s": 250.624329, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-d1eb1d4d-a0a6-42e6-964d-1c22a0266886", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/MobileNet-v3-large:mobilenet_v3_large.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/MobileNet-v3-large", "file": "mobilenet_v3_large.tflite", "task": "image-classification", "model_size_mb": 21.9215832, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 134, "nodes_total": 134, "partitions": 1, "latency_ms": {"median": 0.893, "avg": 0.895976722, "p5": 0.862, "p95": 0.941, "min": 0.827, "max": 1.163, "stddev": 0.026, "init": 672.977, "first_inference": 5.849, "warm_up_avg": 0.991604}, "memory_mb": {"init_footprint": 44.25, "overall_footprint": 76.23}, "runs": 1117, "warmup_runs": 505, "throughput_mb_s": 640.885803, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-64430cf4-a741-4c75-9514-14e80de7c1d2", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/efficientnet_b0:efficientnet_b0.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/efficientnet_b0", "file": "efficientnet_b0.tflite", "task": "image-classification", "model_size_mb": 21.1264477, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 244, "nodes_total": 244, "partitions": 1, "latency_ms": {"median": 6.984, "avg": 6.99117374, "p5": 6.95, "p95": 7.056, "min": 6.939, "max": 7.221, "stddev": 0.039, "init": 38.314, "first_inference": 11.682, "warm_up_avg": 7.2297}, "memory_mb": {"init_footprint": 22.53, "overall_footprint": 34.11}, "runs": 144, "warmup_runs": 70, "throughput_mb_s": 82.1348114, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-ad739235-0889-4c06-a41e-cd9408a6cc23", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/efficientnet_b0:efficientnet_b0.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/efficientnet_b0", "file": "efficientnet_b0.tflite", "task": "image-classification", "model_size_mb": 21.1264477, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 244, "nodes_total": 244, "partitions": 1, "latency_ms": {"median": 1.301, "avg": 1.30221617, "p5": 1.281, "p95": 1.327, "min": 1.259, "max": 1.372, "stddev": 0.014, "init": 517.367, "first_inference": 7.039, "warm_up_avg": 1.42113924}, "memory_mb": {"init_footprint": 44.16, "overall_footprint": 77.28}, "runs": 768, "warmup_runs": 352, "throughput_mb_s": 440.955017, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-d1959f26-c2ea-4fb2-b585-dd380ca72416", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/efficientnet_v2_s:efficientnet_v2_s.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/efficientnet_v2_s", "file": "efficientnet_v2_s.tflite", "task": "image-classification", "model_size_mb": 85.6423492, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 508, "nodes_total": 508, "partitions": 1, "latency_ms": {"median": 41.452, "avg": 42.0171585, "p5": 40.186, "p95": 44.887, "min": 39.978, "max": 45.367, "stddev": 1.314, "init": 185.863, "first_inference": 50.109, "warm_up_avg": 40.3006935}, "memory_mb": {"init_footprint": 87.19, "overall_footprint": 119.25}, "runs": 50, "warmup_runs": 13, "throughput_mb_s": 40.1621628, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-b9590e97-7d4b-4140-a208-e89b3b38de5e", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/efficientnet_v2_s:efficientnet_v2_s.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/efficientnet_v2_s", "file": "efficientnet_v2_s.tflite", "task": "image-classification", "model_size_mb": 85.6423492, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 508, "nodes_total": 508, "partitions": 1, "latency_ms": {"median": 10.643, "avg": 10.6677341, "p5": 10.554, "p95": 10.787, "min": 10.531, "max": 10.801, "stddev": 0.084, "init": 1100.97, "first_inference": 27.137, "warm_up_avg": 10.8545532}, "memory_mb": {"init_footprint": 180.98, "overall_footprint": 215.44}, "runs": 94, "warmup_runs": 47, "throughput_mb_s": 158.187302, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-d771ce13-fe0b-4863-8588-06b8676cd5e4", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/resnet18:resnet18.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/resnet18", "file": "resnet18.tflite", "task": "image-classification", "model_size_mb": 46.7505608, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 37, "nodes_total": 37, "partitions": 1, "latency_ms": {"median": 4.114, "avg": 4.11783123, "p5": 3.995, "p95": 4.252, "min": 3.979, "max": 4.348, "stddev": 0.088, "init": 124.254, "first_inference": 9.592, "warm_up_avg": 4.18571663}, "memory_mb": {"init_footprint": 45.17, "overall_footprint": 58.64}, "runs": 243, "warmup_runs": 120, "throughput_mb_s": 139.446884, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-16a094ac-2fcd-40fa-b680-fdf3086288be", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/resnet18:resnet18.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/resnet18", "file": "resnet18.tflite", "task": "image-classification", "model_size_mb": 46.7505608, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 37, "nodes_total": 37, "partitions": 1, "latency_ms": {"median": 2.351, "avg": 2.35486579, "p5": 2.332, "p95": 2.387, "min": 2.317, "max": 2.437, "stddev": 0.017, "init": 551.663, "first_inference": 11.791, "warm_up_avg": 2.44585848}, "memory_mb": {"init_footprint": 100.44, "overall_footprint": 133.02}, "runs": 425, "warmup_runs": 205, "throughput_mb_s": 243.843506, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-a58676ba-01f7-4ad5-a1dc-669de9aa0dd9", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/squeezenet1_1:squeezenet1_1.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/squeezenet1_1", "file": "squeezenet1_1.tflite", "task": "image-classification", "model_size_mb": 4.95489597, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 40, "nodes_total": 40, "partitions": 1, "latency_ms": {"median": 0.98, "avg": 0.981424, "p5": 0.978, "p95": 0.989, "min": 0.976, "max": 1.027, "stddev": 0.004, "init": 4.283, "first_inference": 2.159, "warm_up_avg": 1.00851214}, "memory_mb": {"init_footprint": 5.11, "overall_footprint": 11.3}, "runs": 1019, "warmup_runs": 496, "throughput_mb_s": 585.087341, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-82b4dace-8f9e-406e-b1bf-95c3adbe2526", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/squeezenet1_1:squeezenet1_1.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/squeezenet1_1", "file": "squeezenet1_1.tflite", "task": "image-classification", "model_size_mb": 4.95489597, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 40, "nodes_total": 40, "partitions": 1, "latency_ms": {"median": 0.793, "avg": 0.795059621, "p5": 0.786, "p95": 0.813, "min": 0.774, "max": 0.858, "stddev": 0.008, "init": 321.063, "first_inference": 5.018, "warm_up_avg": 0.895288}, "memory_mb": {"init_footprint": 12.62, "overall_footprint": 42.67}, "runs": 1258, "warmup_runs": 559, "throughput_mb_s": 722.233582, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-f556b164-8f1b-41ee-b01b-1fe03e31e3ec", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/shufflenet_v2_x0_5:shufflenet_v2_x0_5.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/shufflenet_v2_x0_5", "file": "shufflenet_v2_x0_5.tflite", "task": "image-classification", "model_size_mb": 5.64969206, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 189, "nodes_total": 190, "partitions": 3, "latency_ms": {"median": 0.596, "avg": 0.601316273, "p5": 0.572, "p95": 0.661, "min": 0.568, "max": 0.684, "stddev": 0.026, "init": 3.297, "first_inference": 1.389, "warm_up_avg": 0.601265609}, "memory_mb": {"init_footprint": 6.45, "overall_footprint": 11.25}, "runs": 1663, "warmup_runs": 832, "throughput_mb_s": 954.936279, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-49738ce6-9a49-4192-bc43-c2333a3c5860", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/shufflenet_v2_x0_5:shufflenet_v2_x0_5.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/shufflenet_v2_x0_5", "file": "shufflenet_v2_x0_5.tflite", "task": "image-classification", "model_size_mb": 5.64969206, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 10, "nodes_total": 190, "partitions": 3, "latency_ms": {"median": 0.98, "avg": 0.977666676, "p5": 0.95, "p95": 0.994, "min": 0.935, "max": 1.045, "stddev": 0.013, "init": 155.028, "first_inference": 3.748, "warm_up_avg": 1.03917217}, "memory_mb": {"init_footprint": 8.5, "overall_footprint": 42.44}, "runs": 1023, "warmup_runs": 482, "throughput_mb_s": 587.335938, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-bbe66c98-ca6d-4947-bb5a-a99d1e7f3869", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/googlenet:googlenet.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/googlenet", "file": "googlenet.tflite", "task": "image-classification", "model_size_mb": 26.5056, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 93, "nodes_total": 105, "partitions": 25, "latency_ms": {"median": 4.159, "avg": 4.22259903, "p5": 3.756, "p95": 4.728, "min": 3.718, "max": 4.913, "stddev": 0.281, "init": 50.365, "first_inference": 7.663, "warm_up_avg": 4.04284668}, "memory_mb": {"init_footprint": 26.67, "overall_footprint": 41.81}, "runs": 237, "warmup_runs": 124, "throughput_mb_s": 135.98703, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-928c3f9a-f7b8-48f2-b375-0607a9b59fe8", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/googlenet:googlenet.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/googlenet", "file": "googlenet.tflite", "task": "image-classification", "model_size_mb": 26.5056, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 105, "nodes_total": 105, "partitions": 1, "latency_ms": {"median": 2.464, "avg": 2.47318029, "p5": 2.431, "p95": 2.522, "min": 2.404, "max": 3.617, "stddev": 0.063, "init": 583.606, "first_inference": 11.227, "warm_up_avg": 2.54112697}, "memory_mb": {"init_footprint": 65.58, "overall_footprint": 99.52}, "runs": 405, "warmup_runs": 197, "throughput_mb_s": 232.178284, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-d27f1ffa-3d85-4751-8ed1-a825e388ea9c", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/inception_v3:inception_v3.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/inception_v3", "file": "inception_v3.tflite", "task": "image-classification", "model_size_mb": 95.5243, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 139, "nodes_total": 141, "partitions": 5, "latency_ms": {"median": 11.329, "avg": 11.4112272, "p5": 11.224, "p95": 12.05, "min": 11.188, "max": 12.303, "stddev": 0.236, "init": 53.807, "first_inference": 21.239, "warm_up_avg": 11.9050951}, "memory_mb": {"init_footprint": 93.64, "overall_footprint": 110.83}, "runs": 88, "warmup_runs": 42, "throughput_mb_s": 89.6584778, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-153f156e-a031-4cde-b8a5-e492d59da1be", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/inception_v3:inception_v3.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/inception_v3", "file": "inception_v3.tflite", "task": "image-classification", "model_size_mb": 95.5243, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 141, "nodes_total": 141, "partitions": 1, "latency_ms": {"median": 8.048, "avg": 8.0564, "p5": 7.985, "p95": 8.156, "min": 7.953, "max": 8.269, "stddev": 0.058, "init": 1206.80603, "first_inference": 21.256, "warm_up_avg": 8.24901676}, "memory_mb": {"init_footprint": 198.97, "overall_footprint": 233.89}, "runs": 125, "warmup_runs": 61, "throughput_mb_s": 126.993851, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-51f769b1-8afd-4868-aadd-e4e4c729e77e", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/convnext_tiny:convnext_tiny.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/convnext_tiny", "file": "convnext_tiny.tflite", "task": "image-classification", "model_size_mb": 114.422653, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 160, "nodes_total": 183, "partitions": 47, "latency_ms": {"median": 57.001, "avg": 57.9995804, "p5": 55.494, "p95": 61.93, "min": 55.036, "max": 62.766, "stddev": 2.293, "init": 200.02, "first_inference": 64.239, "warm_up_avg": 56.0164452}, "memory_mb": {"init_footprint": 112.31, "overall_footprint": 133.14}, "runs": 50, "warmup_runs": 9, "throughput_mb_s": 9.90039539, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-f7433517-d0bd-415a-a46e-7c8934ae85f7", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/convnext_tiny:convnext_tiny.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/convnext_tiny", "file": "convnext_tiny.tflite", "task": "image-classification", "model_size_mb": 114.422653, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 183, "nodes_total": 183, "partitions": 1, "latency_ms": {"median": 7.048, "avg": 7.0372448, "p5": 6.958, "p95": 7.08, "min": 6.924, "max": 7.098, "stddev": 0.037, "init": 1041.19, "first_inference": 19.936, "warm_up_avg": 7.11074638}, "memory_mb": {"init_footprint": 170.98, "overall_footprint": 204.41}, "runs": 143, "warmup_runs": 71, "throughput_mb_s": 81.5971, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-8dd402e8-ccac-470e-817b-f2ef39e9e26a", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/vit_base_patch16_224.augreg_in1k:vit_base_patch16_224_fp32.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/vit_base_patch16_224.augreg_in1k", "file": "vit_base_patch16_224_fp32.tflite", "task": "image-classification", "model_size_mb": 346.216125, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 363, "nodes_total": 400, "partitions": 27, "latency_ms": {"median": 49.551, "avg": 50.1644592, "p5": 47.863, "p95": 54.65, "min": 47.777, "max": 55.065, "stddev": 2.185, "init": 571.554, "first_inference": 75.167, "warm_up_avg": 50.9443}, "memory_mb": {"init_footprint": 332.52, "overall_footprint": 358.0}, "runs": 50, "warmup_runs": 10, "throughput_mb_s": 11.4467249, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-70929d44-d77c-482d-b55a-e9d17e9d59b2", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/vit_base_patch16_224.augreg_in1k:vit_base_patch16_224_fp32.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/vit_base_patch16_224.augreg_in1k", "file": "vit_base_patch16_224_fp32.tflite", "task": "image-classification", "model_size_mb": 346.216125, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 28, "nodes_total": 400, "partitions": 3, "latency_ms": {"median": 54.374, "avg": 54.3288803, "p5": 52.205, "p95": 56.597, "min": 52.017, "max": 57.638, "stddev": 1.421, "init": 774.214, "first_inference": 71.68, "warm_up_avg": 54.1054}, "memory_mb": {"init_footprint": 356.24, "overall_footprint": 392.0}, "runs": 50, "warmup_runs": 10, "throughput_mb_s": 10.5693092, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-dde3feba-8241-46dc-83f0-39de92c9b94d", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/lraspp_mobilenet_v3_large:lraspp_mobilenet_v3_large.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/lraspp_mobilenet_v3_large", "file": "lraspp_mobilenet_v3_large.tflite", "task": "image-segmentation", "model_size_mb": 12.8802719, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 156, "nodes_total": 156, "partitions": 1, "latency_ms": {"median": 19.953, "avg": 19.9339027, "p5": 19.425, "p95": 20.435, "min": 19.26, "max": 20.776, "stddev": 0.32, "init": 35.952, "first_inference": 31.976, "warm_up_avg": 20.3373203}, "memory_mb": {"init_footprint": 13.84, "overall_footprint": 81.23}, "runs": 51, "warmup_runs": 25, "throughput_mb_s": 155.237167, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-d7772020-e267-4831-ba4d-5b0aa841a921", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/lraspp_mobilenet_v3_large:lraspp_mobilenet_v3_large.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/lraspp_mobilenet_v3_large", "file": "lraspp_mobilenet_v3_large.tflite", "task": "image-segmentation", "model_size_mb": 12.8802719, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 156, "nodes_total": 156, "partitions": 1, "latency_ms": {"median": 5.498, "avg": 5.50728035, "p5": 5.428, "p95": 5.587, "min": 5.405, "max": 5.637, "stddev": 0.052, "init": 739.684, "first_inference": 18.381, "warm_up_avg": 5.64182043}, "memory_mb": {"init_footprint": 27.38, "overall_footprint": 109.08}, "runs": 182, "warmup_runs": 89, "throughput_mb_s": 561.889404, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-f2929afe-92ee-424a-8208-4c5c953ebf39", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/deeplabv3_resnet50:deeplabv3_resnet50.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/deeplabv3_resnet50", "file": "deeplabv3_resnet50.tflite", "task": "image-segmentation", "model_size_mb": 158.472351, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 86, "nodes_total": 87, "partitions": 3, "latency_ms": {"median": 385.096, "avg": 383.899658, "p5": 342.52, "p95": 418.985, "min": 342.042, "max": 432.683, "stddev": 24.533, "init": 108.837, "first_inference": 399.081, "warm_up_avg": 369.675}, "memory_mb": {"init_footprint": 153.12, "overall_footprint": 376.14}, "runs": 50, "warmup_runs": 2, "throughput_mb_s": 8.06065464, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-dbbd1a8c-3a6a-47ae-9075-c3a10aab1600", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/deeplabv3_resnet50:deeplabv3_resnet50.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/deeplabv3_resnet50", "file": "deeplabv3_resnet50.tflite", "task": "image-segmentation", "model_size_mb": 158.472351, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 87, "nodes_total": 87, "partitions": 1, "latency_ms": {"median": 169.787, "avg": 168.427521, "p5": 161.452, "p95": 171.977, "min": 160.639, "max": 172.097, "stddev": 3.521, "init": 1145.16797, "first_inference": 171.259, "warm_up_avg": 162.509506}, "memory_mb": {"init_footprint": 259.05, "overall_footprint": 352.39}, "runs": 50, "warmup_runs": 4, "throughput_mb_s": 18.3727837, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-7e846428-59ac-4b78-9fec-59655ed22ad8", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/whisper-tiny:whisper_tiny_30s_f32.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/whisper-tiny", "file": "whisper_tiny_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 150.979187, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 73, "nodes_total": 82, "partitions": 18, "latency_ms": {"median": 67.65, "avg": 67.5373, "p5": 67.047, "p95": 67.938, "min": 66.941, "max": 68.074, "stddev": 0.301, "init": 147.624, "first_inference": 84.592, "warm_up_avg": 70.2838745}, "memory_mb": {"init_footprint": 146.25, "overall_footprint": 233.08}, "runs": 50, "warmup_runs": 8, "throughput_mb_s": 13.5558767, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-3afb6145-e438-4219-b477-0d6e94df93b1", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/whisper-tiny:whisper_tiny_30s_f32.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/whisper-tiny", "file": "whisper_tiny_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 150.979187, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 82, "nodes_total": 82, "partitions": 1, "latency_ms": {"median": 27.441, "avg": 27.2811394, "p5": 26.633, "p95": 27.768, "min": 26.587, "max": 28.38, "stddev": 0.434, "init": 1477.65405, "first_inference": 42.024, "warm_up_avg": 27.5486317}, "memory_mb": {"init_footprint": 568.84, "overall_footprint": 629.59}, "runs": 50, "warmup_runs": 19, "throughput_mb_s": 33.5589828, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-7972c6e6-4b90-49dc-bd7a-a12dfa157bd9", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/whisper-base:whisper_base_30s_f32.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/whisper-base", "file": "whisper_base_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 290.082642, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 343, "nodes_total": 343, "partitions": 1, "latency_ms": {"median": 153.639, "avg": 152.041046, "p5": 145.666, "p95": 156.09, "min": 145.445, "max": 156.332, "stddev": 3.494, "init": 751.197, "first_inference": 183.157, "warm_up_avg": 155.588257}, "memory_mb": {"init_footprint": 279.41, "overall_footprint": 370.56}, "runs": 50, "warmup_runs": 4, "throughput_mb_s": 6.02158, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-c0b397db-ff7e-4f65-8445-753704adfc0b", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/whisper-base:whisper_base_30s_f32.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/whisper-base", "file": "whisper_base_30s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 290.082642, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 343, "nodes_total": 343, "partitions": 1, "latency_ms": {"median": 89.552, "avg": 89.6305618, "p5": 88.272, "p95": 90.954, "min": 87.289, "max": 92.279, "stddev": 0.902, "init": 3022.95605, "first_inference": 104.329, "warm_up_avg": 90.5838318}, "memory_mb": {"init_footprint": 662.44, "overall_footprint": 765.42}, "runs": 50, "warmup_runs": 6, "throughput_mb_s": 10.2144556, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-7cf3bd18-759d-42b2-8fec-3c7dc391a754", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/moonshine-tiny:moonshine_tiny_5s_f32.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/moonshine-tiny", "file": "moonshine_tiny_5s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 109.373138, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 321, "nodes_total": 358, "partitions": 38, "latency_ms": {"median": 11.276, "avg": 11.3871822, "p5": 11.209, "p95": 12.028, "min": 11.2, "max": 12.345, "stddev": 0.255, "init": 256.792, "first_inference": 14.52, "warm_up_avg": 11.3021336}, "memory_mb": {"init_footprint": 112.0, "overall_footprint": 120.17}, "runs": 88, "warmup_runs": 45, "throughput_mb_s": 26.7999401, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-91a2ee2b-8b85-4038-aa3b-679e88dab38b", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/moonshine-tiny:moonshine_tiny_5s_f32.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/moonshine-tiny", "file": "moonshine_tiny_5s_f32.tflite", "task": "automatic-speech-recognition", "model_size_mb": 109.373138, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 358, "nodes_total": 358, "partitions": 1, "latency_ms": {"median": 5.988, "avg": 5.98492241, "p5": 5.944, "p95": 6.017, "min": 5.906, "max": 6.061, "stddev": 0.023, "init": 1917.79797, "first_inference": 19.26, "warm_up_avg": 6.11839}, "memory_mb": {"init_footprint": 316.66, "overall_footprint": 346.86}, "runs": 168, "warmup_runs": 82, "throughput_mb_s": 50.9907646, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-2ca58987-ea75-4ff3-84b4-5de397dab2dc", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/Gecko-110m-en:Gecko_256_f32.tflite@2.2.0/ios/iphone16pro-18-3/cpu", "model": "litert-community/Gecko-110m-en", "file": "Gecko_256_f32.tflite", "task": "question-answering", "model_size_mb": 443.197906, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "cpu", "delegate": "TfLiteXNNPackDelegate", "nodes_delegated": 614, "nodes_total": 633, "partitions": 2, "latency_ms": {"median": 48.689, "avg": 48.6158218, "p5": 47.731, "p95": 49.614, "min": 47.409, "max": 50.142, "stddev": 0.557, "init": 585.071, "first_inference": 74.933, "warm_up_avg": 51.5965}, "memory_mb": {"init_footprint": 331.69, "overall_footprint": 360.97}, "runs": 50, "warmup_runs": 10, "throughput_mb_s": 0.020087339, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-fa279c4c-579b-471f-985a-e13051569b74", "job": "cpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}
{"row_id": "litert-community/Gecko-110m-en:Gecko_256_f32.tflite@2.2.0/ios/iphone16pro-18-3/gpu", "model": "litert-community/Gecko-110m-en", "file": "Gecko_256_f32.tflite", "task": "question-answering", "model_size_mb": 443.197906, "platform": "ios", "device_id": "iphone16pro-18-3", "device": "iPhone 16 Pro", "os": "iOS 18.3", "accelerator": "gpu", "delegate": "LITERT_METAL", "nodes_delegated": 589, "nodes_total": 633, "partitions": 2, "latency_ms": {"median": 28.403, "avg": 28.4061, "p5": 28.33, "p95": 28.5, "min": 28.264, "max": 28.606, "stddev": 0.057, "init": 2663.41699, "first_inference": 46.986, "warm_up_avg": 29.4198818}, "memory_mb": {"init_footprint": 301.53, "overall_footprint": 364.92}, "runs": 50, "warmup_runs": 17, "throughput_mb_s": 0.034378618, "source": "results.pb", "runtime": "litert", "runtime_version": "2.2.0", "binary": "source v2.2.0 (145c7523f), XCTest (Release) in benchmark/ios", "runner": "ddp", "session": "session-5de7eedd-780f-4bbe-82d7-6625509a6fe6", "job": "gpu-iphone16pro-18-3", "status": "measured", "date": "2026-10-03"}