FastFrames
Apple video models that make 480p from text almost in real time: autoregressive diffusion in just two steps. The 14B L version is built on Wan2.1-T2V-14B, and the S version is 1.3B. A 16-times faster version of LingBot 1.3B with camera control is published alongside, and an interactive demo is built on it. Linux, an NVIDIA graphics card and CUDA 12.8 are needed. The license is for research: it can be used only for scientific purposes, not in products or for orders.
- Developer
- Apple, USA
- First release
- Nov 2026
- Latest release
- Nov 2026
- Sizes
- 1.3B (S) and 14B (L)
- License
- Non-commercial onlyApple Machine Learning Research Model License: for scientific research only, commercial use is not allowed
- Running
- On your own serverNeeds a GPU
- Industries
- Media and production, Science and research
What it does
- research on fast video generation
- interactive prototypes with a controllable camera
- comparing speed and quality with other video models
Where it is used
Hardware requirements
Versions
- FastFrames-L 14B
- FastFrames-S 1.3B
How to run it
I can set this up end to end: pick the model size, deploy it on your server and connect it to your systems.
Frequently asked questions
Can FastFrames be used in a commercial project?
No. License: Apple Machine Learning Research Model License: for scientific research only, commercial use is not allowed. A commercial product needs a different model or a separate agreement with the rights holder.
What hardware does FastFrames need?
At minimum: One GPU with 16–80 GB — mid-size versions. Without a GPU the model is not practical. You can calculate the exact VRAM for your model size and context in the hardware calculator.
Does FastFrames support Russian?
Language does not matter for this model: it does not work with text.
Where can I download FastFrames and what does it cost?
The FastFrames weights are open and free to download. You only pay for the hardware it runs on and for the setup. Source links are at the bottom of this page.
How I deploy it for clients
- SelectionI pick the model size for your task and hardware and test it on your examples.
- DeploymentI deploy it on your server or in a closed network and provide an API.
- Fine-tuningI fine-tune it on your data (LoRA) or connect a knowledge base — whichever is cheaper for the task.
- IntegrationI connect it to your CRM, ERP, bot, website or team chat and set up monitoring.
Similar models
Text-to-video and image-to-video; the small version runs on a gaming GPU. After 2.2 only applied models are open: editing (VACE), audio-driven talking characters (S2V), dancing to music (Dancer).
DetailsVideoHunyuanVideoTencent · ChinaCommercial use with conditionsTencent's video model, one of the first open ones on par with closed services. Version 1.5 is lighter (8.3B) and runs on consumer GPUs.
DetailsVideoLongCat-VideoMeituan · ChinaCommercial use allowedA 13.6B video model: from text, from an image and video continuation. Keeps quality on clips several minutes long.
DetailsSource: huggingface.co/apple/FastFrames-L. Data checked against the model card on 22 Sep 2026. Have a lawyer review the license before commercial launch.


