Mistral launches Voxtral 3: SOTA ASR, TTS and voice-agent SDK, top of AA-WER charts
sivareddyg · x · 2026-10-21
Mistral released Voxtral 3, a family of speech models: Mini & Small ASR for batch and real-time streaming transcription, expressive multilingual TTS, and a low-latency voice-agent SDK with tool calling — available via API or self-hosted. Artificial Analysis charts show Voxtral 3 Mini/Small at 2.1-2.2% WER (near the top of the non-streaming index) and Voxtral 3 streaming at 3.3%, well ahead of GPT-4o mini Transcribe and Gemini.
Related event: Mistral Launches Voxtral 3 Speech Models and Voice Agents API(10 posts)→
More from Multimodal
- MOSS-Image's text encoder is multimodal and accepts image inputs — RandumbRedditor1000 · 2026-10-21
- Making a classroom case-explainer video with Opus 5.5 — technollama · 2026-10-21
- FastVideo releases FastH3 Omni weight collection with FP8 and NVFP4 ComfyUI builds — multimodalart · 2026-10-21
- Opus 5.5 distills a 30-min pi-durable lecture into a 7-min narrated explainer video — solyarisoftware · 2026-10-21
- MOSS-Image turns out to accept image inputs, opening path to an edit model — RandumbRedditor1000 · 2026-10-21
- Locally-run Minimax H3 generates a Draco Malfoy vs Gordon Ramsay skit video — Riddy_93 · 2026-10-21