Question Answering on NSF-QA Close-talk
68.3Content AccuracyVoxtral Mv1
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Voxtral Mv1Evaluation Protocol=zero-shot2026.06 | 68.3 | 25.5 | 49.7 | |
| Gemini 3.0 FlashEvaluation Protocol=zero-shot2026.06 | 68.1 | 34.1 | 75 |
Loading the SOTA2 catalog…