About Mistral
Our mission is to make frontier AI open to all, and together solve the world's hardest problems.
We partner with organizations in high-stakes industries like finance, manufacturing, defense, energy, and the public sector to co-create tailored AI systems to solve their hardest, most high-value problems. Because we believe that this critical stake for companies requires deep control, ownership, and production reliability, we have built full-stack AI solutions, including frontier models, developer tools, applications, and compute.
Website ↗Documentation ↗
Source ↗ · 2026-11-12T02:17:02.963004760Zmistralai/voxtral-mini-transcribe-2602
Voxtral Mini Transcribe is Mistral's speech-to-text model, derived from the Voxtral Mini family. It accepts audio input and returns transcribed text via the standard transcription API. Suited for transcribing meetings, voice notes, podcasts, and other spoken content.
- Context window
- 16,384
- Input modalities
- audio
- Output modalities
- transcription
- Capabilities
- audioInput
mistralai/voxtral-mini-transcribe-3-0-20261020
Voxtral Mini Transcribe 3 is Mistral's compact speech-to-text model. It takes audio input and returns a transcript with timestamps and speaker diarization. It also supports context biasing for domain terms. It is small enough to self-deploy in resource-constrained environments, a
- Context window
- 65,536
- Input modalities
- audio
- Output modalities
- transcription
- Capabilities
- audioInput, streaming, transcription
mistralai/voxtral-mini-tts-2603
Voxtral Mini TTS is Mistral's text-to-speech model featuring zero-shot voice cloning and multilingual support. It converts text input into natural-sounding audio output.
- Context window
- 4,096
- Input modalities
- text
- Output modalities
- speech
- Capabilities
- audioInput, multimodal, speechOutput, streaming, voiceCloning
mistralai/voxtral-small-24b-2507-stt-20260813
Voxtral Small is an enhancement of Mistral Small 3, incorporating state-of-the-art audio input capabilities while retaining best-in-class text performance. It excels at speech transcription, translation and audio understanding. Input audio is priced at $100 per million seconds.
- Context window
- 0
- Input modalities
- audio
- Output modalities
- transcription
- Capabilities
- audioInput, multimodal
Benchmarks
- Open ASR Leaderboard: 2.47%
Inference prices
- Published price : $0.01 / cache_read_tokens_per_million · openRouter ↗
- Published price : $0.10 / input_tokens_per_million · openRouter ↗
- Published price : $0.30 / output_tokens_per_million · openRouter ↗
mistralai/voxtral-small-transcribe-3-0-20261020
Voxtral Small Transcribe 3 is Mistral's flagship speech-to-text model. It takes audio input and returns a transcript with timestamps and speaker diarization, and supports context biasing for domain-specific vocabulary. Suited for high-accuracy multilingual transcription of meetin
- Context window
- 65,536
- Input modalities
- audio
- Output modalities
- transcription
- Capabilities
- audioInput, streaming, transcription
mistralai/voxtral-mini-tts-3-0-20261020
Voxtral TTS 3 is Mistral's text-to-speech model. It converts text into natural-sounding speech, offers a library of preset voices, supports zero-shot voice cloning, and accepts emotion and pace tags in the input to control delivery. It streams audio with low time-to-first-audio a
- Context window
- 4,096
- Input modalities
- text
- Output modalities
- speech
- Capabilities
- audioInput, multimodal, speechOutput, streaming, voiceCloning
vertexAI/publishers/mistral-ai/models/mixtral@mixtral-8x22b-v0.1
- Input modalities
- text
- Output modalities
- text
Benchmarks
- Open LLM Leaderboard v1: 74.471%
- Open LLM Leaderboard v2: 25.741%
awsBedrock/mistral.mixtral-8x7b-instruct-v0:1
- Input modalities
- text
- Output modalities
- text
vertexAI/publishers/mistral-ai/models/mixtral@mixtral-8x7b-instruct-v0.1
- Input modalities
- text
- Output modalities
- text
Benchmarks
- HELM Capabilities: 67.261%
- LiveBench: 17.493%
- Open LLM Leaderboard v1: 72.705%
- Open LLM Leaderboard v2: 23.817%
vertexAI/publishers/mistral-ai/models/mixtral@mixtral-8x7b-v0.1
- Input modalities
- text
- Output modalities
- text
Benchmarks
- Open LLM Leaderboard v1: 68.472%
- Open LLM Leaderboard v2: 19.665%
mistral/mistral-ocr-2512
Our OCR service powering our Document AI stack that enables our users to extract interleaved text and images. OCR 4 is available as the newer model. OCR 3 remains available for existing integrations and production workloads.
- Input modalities
- image, document
- Output modalities
- text
- Capabilities
- batch, fileInput, multimodal, ocr, structuredOutputs, visionInput
mistral/mistral-ocr-4-1
Our latest OCR service powering our Document AI stack, with native paragraph-level bounding box extraction, structural block labels, and block-level confidence scores.
- Input modalities
- image, document
- Output modalities
- text
- Capabilities
- batch, fileInput, multimodal, ocr, structuredOutputs, visionInput
awsBedrock/mistral.pixtral-large-2502-v1:0
- Input modalities
- text, image
- Output modalities
- text
- Capabilities
- multimodal, visionInput
mistral/shieldstral-1-0
A compact multimodal moderation model for prompt moderation, response moderation, prompt-response pair classification, refusal detection, and safety filtering across text and image inputs. It uses natural-language policy questions and returns a yes or no classification.
- Context window
- 32,000
- Input modalities
- text, image
- Output modalities
- text
- Capabilities
- moderation, multimodal, visionInput
mistral/timestral-1-0-123m
Timestral 1.0 123M is a time series foundation model for zero-shot forecasting. Part of the Timestral family, it offers a balance of forecast accuracy and efficiency, with support for contexts up to 32K timesteps, past-only and known-future covariates, and probabilistic quantile
- Input modalities
- timeseries
- Output modalities
- timeseries
mistral/timestral-1-0-209m
Timestral 1.0 209M is a time series foundation model for zero-shot forecasting. It ranks first among pretrained and zero-shot models on GIFT-Eval, TIME, and fev-bench, with support for contexts up to 32K timesteps, past-only and known-future covariates, and probabilistic quantile
- Input modalities
- timeseries
- Output modalities
- timeseries
mistral/timestral-1-0-28m
Timestral 1.0 28M is a time series foundation model for zero-shot forecasting. Part of the Timestral family, it is optimised for low-latency forecasting, with support for contexts up to 32K timesteps, past-only and known-future covariates, and probabilistic quantile outputs.
- Input modalities
- timeseries
- Output modalities
- timeseries
mistral/timestral-1-0-5m
Timestral 1.0 5M is a time series foundation model for zero-shot forecasting. The smallest model in the Timestral family, it delivers fast, lightweight forecasts, with support for contexts up to 32K timesteps, past-only and known-future covariates, and probabilistic quantile outp
- Input modalities
- timeseries
- Output modalities
- timeseries