Which hardware is fastest for depth estimation?
The fastest result is the Apple M4 (10-core GPU, 120 GB/s) at 133 FPS, running YOLO26n-depth in FP16. That is a documented third-party result, from a page last edited on 2026-10-01. Axelera Metis Embedded 113m M.2 card comes next at 96 FPS, then Axelera Metis Edge 130p PCIe card at 90 FPS. All of these are nano models at batch 1, so they compare directly.
| # | Platform | Result | Power | Perf/W | Price | Grade · date | Source |
|---|---|---|---|---|---|---|---|
| 1 | Apple M4 (10-core GPU, 120 GB/s) | 133 FPS YOLO26n-depth · FP16 | unknown | unknown | unknown | B 2026-10-01 (page last edited) | Ultralytics documentation: Apple Core AI integration benchmark |
| 2 | Axelera Metis Embedded 113m M.2 card | 96 FPS YOLO26n-Depth · INT8 | unknown | unknown | unknown | C 2026-10-15 (page last edited) | Axelera Voyager SDK model zoo (docs/reference/models/model-zoo.md) |
| 3 | Axelera Metis Edge 130p PCIe card | 90 FPS YOLO26n-Depth · INT8 | unknown | unknown | unknown | C 2026-10-15 (page last edited) | Axelera Voyager SDK model zoo (docs/reference/models/model-zoo.md) |
| 4 | Raspberry Pi 5 + AI HAT+ (Hailo-8) | 33.8 FPS YOLO26n-depth · INT8 | unknown | unknown | unknown | B 2026-10-11 (page last edited) | Ultralytics documentation: Hailo integration benchmark |
| 5 | Raspberry Pi 5 (CPU only) | 0.47 FPS YOLO26n-depth · FP32 | unknown | unknown | unknown | B 2026-10-11 (page last edited) | Ultralytics documentation: Hailo integration benchmark |
Filters: task=depth estimation. Power marked “spec” is the vendor's rated figure, not a measurement. Ranking: grade A–C dated results first where available, then headroom and perf/W. Open in /find · JSON
Sources
- https://docs.ultralytics.com/integrations/coreai/
- https://github.com/axelera-ai-hub/voyager-sdk/blob/latest/docs/reference/models/model-zoo.md
- https://docs.ultralytics.com/integrations/hailo/
Generated 2026-10-16 from HWBench data only. All answers · Submit a result
Machine-readable: /find.json (see llms.txt); full tables are in “Show data”. Confidence grades: A audited MLPerf closed · B documented third-party · C community · D vendor-published · E estimate. Unknown values are shown as unknown, never estimated. Generated 2026-10-16 from public sources.