Mistral AIDeepgramAudio matchupSpeech to text
Voxtral Mini Transcribe Realtimevs.Flux
Which model is better?
Voxtral Mini Transcribe Realtime and Flux each lead in different tests.
Pick Voxtral Mini Transcribe Realtime for
Pick Flux for
Shared tests
Independent evaluators and leaderboards only, the newest shared snapshot per evaluator. With 95 % ranges a lead counts when they do not overlap; for hit rates, when it holds in a 95 % test. Without any uncertainty figure at most “slightly ahead”.
Coval
Flux slightly aheadFlux is clearly ahead in 0 of 16 subtests and slightly in 13, Voxtral Mini Transcribe Realtime in 0 clearly and slightly in 3.
Coval Word Error Rate (WER) per-sample p50
lower is better, as of November 5, 2026
Coval Time to First Token (TTFT) maximum
lower is better, as of November 5, 2026
Coval Time to Final Segment (TTFS) mean
lower is better, as of November 5, 2026
Show all 16 subtests
| Subtest | Voxtral Mini Transcribe Realtime | Flux | Verdict |
|---|---|---|---|
| Coval Word Error Rate (WER) per-sample p50 (lower is better) | 0.0 % | 3.1 % | Voxtral Mini Transcribe Realtime slightly ahead |
| Coval Time to First Token (TTFT) maximum (lower is better) | 40,858 ms | 4,616 ms | Flux slightly ahead |
| Coval Time to Final Segment (TTFS) mean (lower is better) | 444 ms | 102 ms | Flux slightly ahead |
| Coval Time to Final Segment (TTFS) p50 (lower is better) | 339 ms | 99 ms | Flux slightly ahead |
| Coval RTF maximum (lower is better) | 15.59 | 5.29 | Flux slightly ahead |
| Coval Time to Final Segment (TTFS) maximum (lower is better) | 35,330 ms | 13,993 ms | Flux slightly ahead |
| Coval AudioToFinal maximum (lower is better) | 43.02 s | 17.25 s | Flux slightly ahead |
| Coval Word Error Rate (WER) per-sample maximum (lower is better) | 191.7 % | 92.3 % | Flux slightly ahead |
| Coval Time to First Token (TTFT) mean (lower is better) | 1,872 ms | 1,095 ms | Flux slightly ahead |
| Coval Time to First Token (TTFT) p50 (lower is better) | 1,846 ms | 1,172 ms | Flux slightly ahead |
| Coval Word Error Rate (WER) per-sample mean (lower is better) | 5.7 % | 6.4 % | Voxtral Mini Transcribe Realtime slightly ahead |
| Coval pooled corpus Word Error Rate (WER) (lower is better) | 5.4 % | 5.7 % | Voxtral Mini Transcribe Realtime slightly ahead |
| Coval RTF mean (lower is better) | 1.07 | 1.02 | Flux slightly ahead |
| Coval AudioToFinal mean (lower is better) | 8.38 s | 8.05 s | Flux slightly ahead |
| Coval RTF p50 (lower is better) | 1.05 | 1.01 | Flux slightly ahead |
| Coval AudioToFinal p50 (lower is better) | 8.70 s | 8.42 s | Flux slightly ahead |
Source · latest November 5, 2026
Vals AI
Flux slightly aheadFlux is clearly ahead in 0 of 3 subtests and slightly in 2, Voxtral Mini Transcribe Realtime in 0 clearly and slightly in 1.
VoiceCodeBench · WER
lower is better, as of October 31, 2026
VoiceCodeBench · Overall
as of October 31, 2026
VoiceCodeBench · CTEM
as of October 31, 2026
Source · latest October 31, 2026
Artificial Analysis
Voxtral Mini Transcribe Realtime slightly aheadArtificial Analysis AA-WER Streaming (final transcription)
lower is better, as of October 15, 2026
Source · latest October 15, 2026
All data side by side
Prices are verbatim from the provider, because audio models bill per minute, character or token.
| Voxtral Mini Transcribe Realtime | Flux | |
|---|---|---|
| Task | Speech to text | Speech to text |
| Developer | Mistral AI | Deepgram |
| Status | Active | Active |
| Weights | open | proprietary |
| Runs locally | yes | no |
| Popularity (out of 100) | 18 | 37 |