Mochi 1
Genmo's Apache-2.0 video model — open weights with strong motion quality, from the Asimov team.
About
Mochi 1 (October 2024, Genmo) is an open text-to-video model known for strong motion quality and prompt adherence — at release it was among the best open models for realistic movement. Built by the Genmo team (makers of the Asimov research program).
Apache 2.0 licensed — commercial use allowed. 8.8k Hugging Face downloads on the preview checkpoint; the checkpoint was updated as recently as October 2026, suggesting ongoing maintenance. A 2024 architecture in a 2026 field — quality trails LTX-2 and the commercial frontier — but the Apache-2.0 terms and motion quality keep it relevant for open video pipelines. Runs via diffusers and ComfyUI.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-claimed at release: leading open motion quality — treat as provider-claimed and dated. No current leaderboard presence.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier). Apache 2.0 with genuinely good motion quality and signs of ongoing maintenance (checkpoint updated Oct 2026). Docked four points: (1) 2024 architecture, trails the current frontier on fidelity; (2) modest adoption (8.8k HF downloads); (3) quality claims are dated provider reports; (4) small team. A respectable open option for motion-focused work.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.1/10
- Maintenance · recency, releases, issue load
- 5.0/10
- Openness · license + docs
- 8.8/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · HF downloads 8,825 (peer-normalized)
- · released 712 days ago
- · manual review: good motion quality; checkpoint updated Oct 2026 (maintenance signal); 2024 architecture trails frontier; license: Apache-2.0