Lumma-Fev-27B
Lumma-Fev-27B is a full-weight fine-tune of Qwen3.8-27B that returns probabilities for typed decision questions instead of generating text. It supports English, Hindi, Tamil and Kannada, and can also accept images as input.
What it does
Use it to score choices or answer structured questions about text or JSON states, including routing requests and assessing urgency.
- Availability
- Available on Hugging Face; use with the lumma-fev package
- Pricing model
- Apache-2.0 license
- Deployment
- Requires a GPU with about 55 GB free, or several smaller GPUs
- Privacy
- Not established
- Licence
- Apache-2.0
Capability
Modality
Access
Deployment
Pricing
Target user
Evidence
6 claims- Returns a probability for every option in one forward pass and does not generate text. [1]
- Works in English, Hindi, Tamil, and Kannada. [1]
It works in **English, Hindi, Tamil and Kannada**,
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - Accepts text, a JSON object, or an array as the state. [1]
`state` can be text, a JSON object or an array.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - Images can be passed with the images argument after installing lumma-fev[images]. [1]
- The model can be loaded with the lumma-fev package. [1]
- It can be served behind a TypeSafe-compatible POST /v1/systemone API. [1]
To serve it behind the TypeSafe-compatible `POST /v1/systemone` API:
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
Sources
1
1
Hugging FaceOwner-controlled
https://huggingface.co/FrontiersMind/Lumma-fev-27B
Retrieved 2026-10-29 00:34 UTC
Verified
Versions
2| Version | Effective | Change | Changed fields | State |
|---|---|---|---|---|
| 2 | 2026-10-15 15:25 UTC | material change | capability, access, pricing, availability, deployment | Current |
| 1 | 2026-10-14 16:22 UTC | publication | None | Current |
In briefs
1- Extract structured text from PDFs and images on a CPU Daily brief 2026-10-14
Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.