Genmo
Makers of Mochi 1, an open-weights text-to-video model
Overview
Genmo builds Mochi 1, a 10-billion-parameter, Apache 2.0-licensed text-to-video model released as a preview in October 2024 and not updated since. Self-host it or use the hosted playground: 250 one-time free credits, then plans from $10 a month; third-party APIs on fal.ai and Replicate run about $0.40 a clip.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 1 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- About 2 Mochi videos in total
- Genmo watermark
- 1,200 credits/month (about 12 Mochi clips)
- High queue priority
- 5,000 credits/month (about 50 Mochi clips)
- Early model access
- Commercial use
Use cases
What you can produce with Genmo
- Mochi 1: 10B-parameter open-weights text-to-video model (Apache 2.0, preview)
- Open-source video autoencoder alongside the model
- Self-hostable via the official GitHub repo or ComfyUI (under 20GB VRAM)
- LoRA fine-tuning on a single 80GB GPU
- Hosted playground with commercial rights on paid plans
- Third-party hosted APIs on fal.ai and Replicate
Genmo's Mochi 1 is a 10-billion-parameter text-to-video model under Apache 2.0, and the licence is still the selling point: fine-tune it, self-host it on your own GPUs through the official repo or ComfyUI, and use the output commercially without asking. It is no longer the frontier of open video, though — Step-Video-T2V and Alibaba's Wan models are larger or newer, the promised HD version never shipped, and the code has not changed since January 2025. Output is short and modest: 480p, up to about five seconds, well behind Sora, Veo or Kling. The hosted playground gives free users roughly two clips in total; paid plans start at $10 a month for about a dozen. There is no first-party API, but fal.ai and Replicate host it. Suitable for researchers who want an open, tunable base model; not a production video tool.
What we found
Genmo released Mochi 1 in October 2024: a 10-billion-parameter text-to-video model under the Apache 2.0 licence, built on what it calls an Asymmetric Diffusion Transformer and released with its video autoencoder. At launch Genmo described it as the largest video model ever openly released, and coverage treated it as an open answer to closed leaders such as Runway and Kling. The company had raised a $28.4 million Series A led by NEA, with The House Fund, Gold House Ventures, WndrCo, Eastlink and Essence VC, and Ion Stoica, Pieter Abbeel and Joey Gonzalez as advisers.
Two years on, the record is thinner than the launch suggested. The release was labelled a preview, and the promised Mochi 1 HD never arrived: Hugging Face still carries only the preview weights, the last code change to the repository was in January 2025, and Genmo's blog has no post after the launch. The open field has also moved past it — Step-Video-T2V, at 30 billion parameters, and Alibaba's Wan 2.1 and 2.2 are all openly licensed. Genmo now describes itself as a research lab working on world models and is hiring, so it is active, but not shipping publicly.
The model itself remains useful for its licence. Output is 480p at 30 frames per second for up to 5.4 seconds, tuned for photorealism rather than animation. The reference code needs about 60 GB of GPU memory, ComfyUI runs it in under 20 GB, and a LoRA trainer fine-tunes it on one 80 GB H100 or A100. The quality claims at launch — prompt adherence judged by Gemini 1.5 Pro, motion scored by Elo from Genmo's own raters — are vendor evaluations.
The hosted playground is small: 250 credits in total for free users, unlocked by adding a card, with a Mochi clip costing 100. Lite costs $10 a month for 1,200 credits and Standard $30 for 5,000, both watermark-free with commercial use. Genmo offers no API of its own; fal.ai and Replicate host Mochi at about 40 cents a clip.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including Genmo's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Genmo unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Genmo directly →
Spotted something out of date? Suggest an update →
Alternatives to Genmo
More in Video & Audio
