Voxtral 3 Release - Voice Code Bench - Overall: leaderboard
Voxtral 3 release evaluation (2026-10-20): exact chart marks retain model, series and score. FLEURS 45 macro-averages WER, using CER for Japanese, Korean, Thai and Chinese; language forced/inferred runs stay separate. Diarization averages cpWER over AMI-IHM, AMI-SDM, Switchboard, CallHome and AISHELL-4 with 30-minute clips. Context-bias recall uses Earnings22. Realtime latency configurations remain distinct. TTS tests retain their release-specific populations and judge; chart x-axis prices are never scores. Faithful Clips (%).
Metric: Faithful Clips (%). Source: mistral.ai. Saturation forecast: Rough model projection: around 2026. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Eleven v4 | 98.33 | |
| 2 | Eleven v4 Turbo | 97.5 | |
| 3 | Voxtral Mini TTS 3.0 | 79.67 | |
| 4 | Inworld TTS-2 | 77.33 | |
| 5 | Inworld TTS-2 Flash | 75 | |
| 6 | Fish Drama 3 | 72.83 | |
| 7 | Gemini 3.8 Flash TTS | 60.33 | |
| 8 | Qwen-Audio 3.1 Flash | 58.67 | |
| 9 | Qwen-Audio 3.0 Plus | 55.83 | |
| 10 | Cartesia Sonic 3.6 | 50.83 |
Interactive version: theaggregate.ai/benchmark?slug=voxtral-3-release-voice-code-bench-overall · How It Works · Data refreshed daily, snapshot 2026-10-21.