FastFrames Models by Apple Machine Learning Research
Apple Machine Learning Research releases · 2 in the last 12 monthsChangelog →
2 models2026
Last refreshed 2026-11-18. Next refresh: weekly.
Capabilities
VisionAll models
MultimodalAll models
Links
WebsiteAbout
Apple’s autoregressive video diffusion models for interactive generation.
Decision facts
- Best fit
- video generationvision and multimodal workcoding
- Capability starting point
- FastFrames L with multimodal inputs
- Lowest tracked input
- Not tracked
- Closest related family
- Apple Intelligence On-Device
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
2 in view
FastFrames SCurrent
Use when the workload needs video generation, 1.3B parameters, and multimodal inputs.
2026-11video generation1.3B parametersmultimodal inputs
FastFrames LCurrent
Use when the workload needs video generation, 14B parameters, and multimodal inputs.
2026-11video generation14B parametersmultimodal inputs
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| FastFrames S | Use when the workload needs video generation, 1.3B parameters, and multimodal inputs. | 2026-11 | video generation1.3B parametersmultimodal inputs | Current |
| FastFrames L | Use when the workload needs video generation, 14B parameters, and multimodal inputs. | 2026-11 | video generation14B parametersmultimodal inputs | Current |
Release Timeline
1 release group2026-11
2 current
FastFrames L
Currentvideo generation14B parametersmultimodal inputs
FastFrames S
Currentvideo generation1.3B parametersmultimodal inputs
Specifications(2 models)
| Model | Released | Parameters | Vision | Multimodal |
|---|---|---|---|---|
| FastFrames S | 2026-11 | 1.3B | Yes | Yes |
| FastFrames L | 2026-11 | 14B | Yes | Yes |



