The official article, model cards and code repository confirm the 4B/27B weights, license, runtime requirements and publisher-run evaluation scope.
FrontiersMind releases Lumma-Fev-v2 decision models
2026-10-30Verified announcementTRIAL
FrontiersMind released ungated Apache-2.0 weights for Lumma-Fev-v2 at 4B and 27B, plus public inference code. The models target risk-sensitive decisions rather than general chat. The 27B model requires lumma-fev 0.3 or newer and about 55 GB of free GPU memory; smaller 0.1B and 0.6B variants remain announced but unavailable.
Availability: Ungated 4B and 27B weights plus public code; 0.1B and 0.6B announced but not released
What this means for you
For teams building constrained decision systems, public weights make a local comparison possible. The reported accuracy and calibration come from the publisher, including a 400-row Indic set translated with Claude, so they do not establish production reliability.
Evidence & related material
Technical details & scores
| Availability | Ungated 4B and 27B weights plus public code; 0.1B and 0.6B announced but not released |
|---|---|
| License | Apache-2.0 |
| Runtime | 27B requires lumma-fev>=0.3, about 55 GB free GPU memory and 8,192-token input |
| Vendor Evidence | Publisher reports 87.3% and 82.4% on selected decision tasks; no independent reproduction found |
| Time Basis | Hugging Face feed timestamp 2026-10-29T16:12:19.790Z, inside the scan window |
Technical impact 3
Engineering value 3
Adoption signal 2
Maturity 2
Verification cost 3
Risk 3