All articles
LLM API Comparison & Cost/

Voxtral TTS 3 API Price and Benchmarks

Official API costs, performance metrics, and voice cloning comparisons against ElevenLabs.

1 minute read

Voxtral TTS 3 API Price and Benchmarks. A macro shot of a precision-cut silicon wafer reflecting amber server-rack LEDs, resting on a brushed aluminum heatsink laced with condensation droplets from a closed-loop liquid cooling block — GPUniq
Data as of 5 sources

Drafted with AI tools and edited by a human. The figures and conclusions were checked by a GPUniq editor.

TL;DR

Mistral AI prices Voxtral TTS 3 at $13.0 per 1 million characters. The model scores a 68.4% voice cloning win-rate against ElevenLabs Flash v2.5. It supports zero-shot cloning, streaming, and expressiveness controls.

API Pricing

Mistral AI prices text-to-speech output at $13.0 per 1 million characters. Self-hosting is unavailable because model weights are not published.

Model EndpointAPI Cost, $/1M charsTarget WorkloadAvailability
voxtral-mini-tts-3-013.0Text-to-speech outputPublic Preview

Voice Cloning Benchmarks

The model achieves a 68.4% voice cloning win-rate against ElevenLabs Flash v2.5. This score comes from Mistral AI's documentation. The exact measurement methodology is not specified. Pre-built stock voice profiles hit a 58.3% naturalness win-rate in the same evaluation.

Multilingual And Expressive Scores

Multilingual judge scores reach 3.16 out of 5 on BenchLM. Expressive English generation scores 3.13 out of 5. Official documentation lists 9 supported languages, while launch marketing materials cite over 35 languages and 50 dialects. Pronunciation control accuracy stands at 86.8%.

Latency And Controls

Time-to-first-audio latency drops to approximately 90 ms when using streaming mode. The API endpoint voxtral-mini-tts-3-0 supports batch and streaming requests. Developers get access to 34 emotion tags and 2 pace tags for fine-grained generation control.

Methodology

Facts are compiled from the vendor's own announcement and documentation, linked below, at the time of writing; where the vendor hasn't published a number it is marked as such. GPUniq prices, when the model is already available here, come from our live catalog.

Sources

  1. 1.Voxtral TTS 3 — Model Card — Mistral Docs — Mistral AI (accessed )
  2. 2.Changelog — Mistral Docs (Voxtral 3 release) — Mistral AI (accessed )
  3. 3.Voxtral 3: Voice Agents SDK & Speech Models — Mistral AI (accessed )
  4. 4.Voxtral TTS 3 Context, Specs & Sources (BenchLM.ai profile) — BenchLM (accessed )
  5. 5.Voxtral Mini TTS 3 Benchmarks and Pricing — Van Data Team — Van Data Team (accessed )

Want cheap GPUs for your next project?

Browse live GPU prices and rent the right card in seconds — H100, A100, RTX 4090, and 50+ more models.

More from the GPUniq blog