Tier 2 Restricted open-weight
What the publisher claims: open source, open weights
The one-liner: Open except for the checkpoints you actually want.
The six questions
All checkpoints downloadable.
Only Small is Apache 2.0; the useful larger variants are non-commercial.
Standard tooling.
Synthetic-data training pipeline described in detail; pseudo-label corpus not shipped.
Pipeline depends on proprietary teacher models and large-scale data.
The HKU-led DepthAnything repository is openly maintained with outside contributors recognised in the tracker and merged pull requests, but the model design and checkpoint roadmap follow the academic team's own plan.
License
Mixed: Small = Apache 2.0; Base/Large/Giant = CC-BY-NC 4.0 — license source ↗
- Non-commercial on all but the Small variant
What is actually released
- code
- four checkpoint sizes with split licensing
- metric-depth fine-tunes
Why it matters
A permissiveness gradient inside a single release: the smallest model is free, the capable ones are not.
Sources
- https://github.com/DepthAnything/Depth-Anything-V2
- https://huggingface.co/depth-anything/Depth-Anything-V2-Small-hf
- https://huggingface.co/depth-anything/Depth-Anything-V2-Large-hf
Derived from the openness dataset behind But Is It Open? — snapshot 2026-11-07 ↗. Scores are qualitative readings of the cited evidence, not verdicts; tier measures openness, not intelligence.