Voxtral-Mini-3B-2507: benchmark results

Provider: Other. Released 2025-07-01. Access: Open.

Unified ELO 1411 ± 17, rank #1114 of 1605 rated models, from 189 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
SEA-SpeechBench - Speech Translation (English Prompt)19.98BLEU (0-100): corpus BLEU of the English translation of Sout92.9
EuroEval Finnish NLU - Tydiqa FI73.4Reading comprehension Score (%)91.2
SEA-SpeechBench - Age Recognition (English Prompt)37.26Macro-F1 (%; speaker age group (teens, adults, seniors) from85.7
EuroEval English NLU - SQuAD84.74Reading comprehension Score (%)85.6
EuroEval German NLU - Germanquad63.91Reading comprehension Score (%)85.6
EuroEval Lithuanian NLU - MultiWikiQA LT67.8Reading comprehension Score (%)85.6
EuroEval Icelandic NLU - NQII54.94Reading comprehension Score (%)84.9
EuroEval Spanish NLU - MLQA ES65.46Reading comprehension Score (%)84.6
EuroEval Polish NLU - PoQuAD55.15Reading comprehension Score (%)81.5
EuroEval French NLU - FQuAD70.85Reading comprehension Score (%)80.4
EuroEval Latvian NLU - MultiWikiQA LV64.57Reading comprehension Score (%)78.9
SEA-SpeechBench - Speech Translation (SEA Prompt)18.15BLEU (0-100): corpus BLEU of the English translation of Sout78.6

Interactive version: theaggregate.ai/model?slug=voxtral-mini-3b-2507 · How It Works · Data refreshed daily, snapshot 2026-09-26.