Our models.
Explore the HeyBlinki model family. Starting with motion.
Blinki Motion
Video generationOur video generation model. The implemented research prototype learns procedural moving shapes; cinematic generation is not yet established.
Blinki Motion 0.0 — Research
2.38M parameters · HeyBlinki
- Owner
- HeyBlinki
- Parameters
- 2,384,371 (2.38M parameters)
- Resolution
- 32 × 32 pixels
- Clip
- 1 second · 8 frames at 8 fps
- Audio
- None
- Availability
- Research package + developer API on configured hosts
Training
Prototype run completed; 12,000 generator steps and 3,000 codec steps. The prototype learns from our procedural moving-shape videos.
What exists today
An implemented research model that generates small, one-second clips. It is not yet a cinematic generator. Try its research API in the developer console on a configured host; Chat and Compose generation remain unavailable.
Blinki Motion 0.1 — Training
The next version is under development. Specifications and results will be published when verified.
The rest of the family
Future owned models. These are not implemented or available for generation.
Blinki Image
Image generation
Blinki Language
Language and planning
Blinki Voice
Speech generation
Blinki Audio
Music and audio generation
HeyBlinki uses owned models only. App generation will become available as our models are ready.
Developer consoleWorkspace documentation