Vega 4B: public intents
Vega 4B is a locally runnable decision model with a frozen Qwen3.5-4B backbone and a trained decision engine. It supports text and image inputs, typed choice, boolean, and score answers, and a context length of 73,728 tokens.
What it does
Use it to make calibrated decisions from text or images, including decisions based on long documents, on your own hardware.
- Availability
- 4B checkpoint under the repository's 4b/ directory; public checkpoints load anonymously
- Pricing model
- No per-decision fee
- Deployment
- Running on your own hardware
- Privacy
- The announcement says local models run with nothing leaving the machine
- Licence
- Apache-2.0
Capability
Modality
Access
Deployment
Pricing
Target user
Evidence
5 claims- The model supports images through the frozen backbone's vision encoder. [1]
| **images** | yes, the frozen backbone's vision encoder, via `vega_vision.py` |
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The 4B checkpoint is available under the repository's 4b/ directory. [1]
- Vega 4B supports choice, boolean, and score answer types. [1]
| answer types | `choice`, `boolean` (probability a statement is true), `score` |
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The stated maximum context length is 73,728 tokens. [1]
- The announcement says it runs on the user's own hardware. [1]
A different and narrower measurement, from before the menu above existed: our own runs of public
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
Sources
1
1
Hugging FaceOwner-controlled
https://huggingface.co/nandakishorm/vega-08b-public-intents
Retrieved 2026-10-28 00:51 UTC
Verified
Versions
3| Version | Effective | Change | Changed fields | State |
|---|---|---|---|---|
| 3 | 2026-10-12 14:02 UTC | evidence update | score | Current |
| 2 | 2026-10-11 14:55 UTC | material change | capability, access, pricing, availability, privacy | Current |
| 1 | 2026-10-10 15:50 UTC | publication | None | Current |
In briefs
1- Two local models can classify, route, or score text and images Daily brief 2026-10-10
Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.