HWBench

Find edge AI hardware

JSON

Task depth estimation · model-size class nano (YOLO-n) · batch 1 · 5 platforms

Not like-for-like across groups: same model-size class and batch, but 4 different model/precision combinations. Compare within a group; each group is sorted by result.

YOLO26n-depth · FP16 (1)

PlatformResultModel · precision · stackGrade · datePowerPriceSource
Apple M4 (10-core GPU, 120 GB/s)133 FPS
latency_e2e b=1
YOLO26n-depth
FP16 · Ultralytics 8.4.168 Core AI (coreai-torch 0.4.3), macOS 27.0
B 2026-10-01 (page last edited)unknown
no power published
unknownUltralytics documentation: Apple Core AI integration benchmark

YOLO26n-Depth · INT8 (2)

PlatformResultModel · precision · stackGrade · datePowerPriceSource
Axelera Metis Embedded 113m M.2 card96.00 FPS
throughput b=1
YOLO26n-Depth
INT8 · Voyager SDK (GStreamer pipeline)
C 2026-10-15 (page last edited)unknown
no power published
unknownAxelera Voyager SDK model zoo (docs/reference/models/model-zoo.md)
Axelera Metis Edge 130p PCIe card90.00 FPS
throughput b=1
YOLO26n-Depth
INT8 · Voyager SDK (GStreamer pipeline)
C 2026-10-15 (page last edited)unknown
no power published
unknownAxelera Voyager SDK model zoo (docs/reference/models/model-zoo.md)

YOLO26n-depth · INT8 (1)

PlatformResultModel · precision · stackGrade · datePowerPriceSource
Raspberry Pi 5 + AI HAT+ (Hailo-8)33.78 FPS
inference_time b=1
YOLO26n-depth
INT8 · Ultralytics Hailo HEF / HailoRT
B 2026-10-11 (page last edited)unknown
no power published
unknownUltralytics documentation: Hailo integration benchmark

YOLO26n-depth · FP32 (1)

PlatformResultModel · precision · stackGrade · datePowerPriceSource
Raspberry Pi 5 (CPU only)0.470 FPS
inference_time b=1
YOLO26n-depth
FP32 · Ultralytics PyTorch (CPU)
B 2026-10-11 (page last edited)unknown
no power published
unknownUltralytics documentation: Hailo integration benchmark

Confidence: A audited MLPerf closed, B documented third-party, C community, D vendor-published, E estimate. Unknown is never assumed. Other like-for-like groups: nano (YOLO-n), batch 1 (5); small (YOLO-s), batch 1 (2); medium (YOLO-m), batch 1 (2); large (YOLO-l), batch 1 (2); xlarge (YOLO-x), batch 1 (2); small ViT, batch 1 (1); unknown, batch 1 (4); unknown, batch not stated (5)