Tier 3 Practical open-weight
What the publisher claims: open-source, open weights
The one-liner: The video model whose license file says Apache-2.0 and stops there.
The six questions
Weights are downloadable from the publisher's repositories without gating.
Apache-2.0 with no revenue threshold, no territory limits, and no field-of-use carve-out, unlike neighbouring video releases.
Runs locally on chosen hardware with published inference code and community serving stacks.
Architecture and capability are described; the training corpus and recipe are not published.
Limited.
Alibaba's Wan 2.2 repository is public and takes issue reports and the occasional outside PR, but the model, its data, and its release cadence are set inside the Wan team.
License
Apache-2.0 OSI-approved — license source ↗
- Standard Apache-2.0 obligations only; no revenue threshold, no territory carve-out, no field-of-use restriction
What is actually released
- weights (T2V-A14B)
- inference code and reference scripts
- model cards
Why it matters
The useful control case for video: identical openness problems to its competitors, but no license caveat attached, which isolates the license variable from the transparency variable.
Sources
- https://github.com/Wan-Video/Wan2.2
- https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B
- https://fuser.studio/articles/best-open-source-video-models
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.