FastFrames-L
Two-step autoregressive video diffusion model for generating 480p frames at interactive latency.
- Arena score
- —Not enough votes yet
- Starting price
- —Not published
- Longest clip
- —Not published
- Resolution
- 480pAudio unconfirmed
Quality & price
How it rates and what it costs
FastFrames-L does not have enough arena votes for a quality score yet.
Specifications
The details
- Lab
- Apple
- Task
- Text-to-video generation
- Published capabilities
- Two-step autoregressive video diffusion model for generating 480p frames at interactive latency.
- Model size
- 14B parameters
- Access
- Downloadable checkpoints with Apple inference code
- Operating limits
- Reference implementation renders 832 × 480 video at 16 fps, with 81 frames in the published example.
- Sources
- Publisher documentation · Model weights
- Type
- Video model
- Released
- Nov 2026
- Inputs
- Text prompts
- Output
- Video
- Can do
- Text to video
- Resolutions
- 480p
- Weights
- Open weights
- Availability
- Not on the API hosts we track
Same lab
More video models from Apple
Questions
About FastFrames-L
Who makes FastFrames-L?
FastFrames-L is a video model from Apple, released Nov 2026.
What can FastFrames-L do?
FastFrames-L accepts text prompts and produces video. In practice: text to video.
Is FastFrames-L open source?
Yes. Apple publishes the weights, so you can run it on your own hardware.
Where can I use FastFrames-L?
Downloadable checkpoints with Apple inference code
What are the alternatives to FastFrames-L?
The closest video models from other labs, by release date, are FastFrames-S (Apple, Nov 2026), Viggle Remake (Viggle AI, Nov 2, 2026), Genie 4 (Google, Oct 27, 2026), PAN 2 (MBZUAI Institute of Foundation Models, Nov 10, 2026) and HeyGen Video (HeyGen, Oct 9, 2026).