Voxtral Small Transcribe 3 is our flagship transcription and diarization model. It achieves SOTA transcription scores, outperforming Microsoft AI Transcribe 2, Whisper large-v3, GPT-Transcribe, Scribe v2, and Gemini 3.5 Transcribe. It also supports translation, speech Q&A, summarization, and function calling.
Voxtral Mini TTS 3 is Mistral's text-to-speech model. It converts text into natural-sounding speech, offers a library of preset voices, and supports zero-shot and professional voice cloning. It supports 35+ languages and 50+ dialects, streaming with ~90 ms time-to-first-audio, 5 output formats, and 34 emotion tags and 2 pace tags in the input to control delivery. It suits voice agents, narration, and multilingual assistants.
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, and supports context biasing for domain terms. It ranks 3rd on the FLEURS multilingual transcription and English diarization benchmarks, ahead of models such as Gemini 3.5 Transcribe, at a competitive price. It is small enough to self-deploy in resource-constrained environments, and fits meeting notes, call analytics, and voice-agent pipelines where cost matters.
Our latest moderation model with 128k context window and jailbreaking detection. Strong performance on complex multilingual data including long multi-turn conversations.
MistralAIOct 30, 2026Voxtral Small 24B 2507 STT is a speech transcription model from Mistral AI. It is suited for transcription, translation, and audio understanding workloads that benefit from its larger model capacity.
Voxtral Mini 3B 2507 is a speech and audio understanding model from Mistral AI. It is suited for transcription, translation, and compact audio processing workloads.
Mistral Large 4 is an open-weight, general-purpose multimodal model (text and image input) with a granular Mixture-of-Experts architecture, built for reasoning, coding, and agentic workloads. It offers a 1M-token context window with up to 256K output tokens, and supports tool calling and structured outputs.
Mistral Small 4 is the next major release in the Mistral Small family, combining the capabilities of several flagship Mistral models into a single system. It brings together strong reasoning from Magistral, multimodal understanding from Pixtral, and agent-oriented coding capabilities from Devstral, allowing one model to handle complex analysis, software development, and visual tasks within the same workflow.
Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total). Released under the Apache 2.0 license.
Mistral Moderation 2411 is a content moderation model from Mistral. It moderates text with high accuracy across nine safety categories and multiple languages, and is designed for real-time use in varied environments.
Ministral 8B is an 8B parameter model with a unique interleaved sliding-window attention pattern for faster and more memory-efficient inference. Optimized for edge use cases, it supports up to 128k context length and delivers strong performance in knowledge and reasoning tasks. Exceeding other models in the sub-10B category, it is ideal for low-latency and privacy-focused applications.
Mistral NeMo is a 12B-parameter model with a 128K-token context window, built by Mistral in collaboration with NVIDIA. It supports English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, and Hindi. It also supports function calling and is released under the Apache 2.0 license.
Codestral is Mistral’s language model for coding, specialized in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction, and test generation. It is built for developer productivity and supports a wide range of programming languages and code-related tasks.
