AutoTrust AI published JEV-35B weights, usage and author-run evaluation.
AutoTrust publishes JEV-35B for one-pass ranking decisions
2026-10-11Frontier watchIGNORE
AutoTrust AI released Apache-2.0 weights for JEV-35B, a Qwen3.5-35B-A3B fine-tune for ranking up to 256 options in one pass. The model card reports a self-scored public benchmark and a 204 ms median on one B200. The full benchmark is mostly private, the result is not a board entry and no independent deployment evidence is available.
What this means for you
One-pass ranking could simplify recommendation pipelines, but a self-scored public subset cannot establish an advantage over established rerankers or task-specific classifiers. The current evidence does not justify a trial.
FRONTIER WATCH
Public artifacts and validated results are different.
The recorded fact is the public work and its artifacts. Author-reported outcomes do not establish independent reproduction. Next checks include usable code, task conditions and third-party validation.
Evidence & related material
Technical details & scores
| Base Model | Qwen3.5-35B-A3B |
|---|---|
| License | Apache-2.0 |
| Task | one-pass ranking or classification over up to 256 options |
| Author Results | Decision Index public subset 59.73; median 204 ms on one B200 |
| Limits | self-scored public subset, not a board entry; private tests dominate the full score |
| Time Basis | model repository created 2026-10-11T01:22:03Z |
Technical impact 2
Engineering value 2
Adoption signal 2
Maturity 2
Verification cost 4
Risk 2