AI provider / Mistral AI
Mistral AI
A provider associated with frontier open-weight models, efficient deployment and a strong European presence across general, coding and multimodal AI.
Mistral AI is Europe’s frontier lab, known for shipping strong open-weight models — often by magnet link — that punch far above their size. From Mistral 7B and the sparse Mixtral to the commercial Large line, it built a reputation for performance with portability.
The catalog now spans general, coding (Codestral), reasoning (Magistral), audio (Voxtral) and edge (Ministral) lines, running both open and commercial tracks. That breadth, with a strong European and infrastructure-flexible story, keeps Mistral on most shortlists.
Model catalog
Every release in this provider’s model line.
Each row is a distinct model release — its date, its type and the traits it is known for. Open a source to dig into any of them.
| Model | Released | Type | Key traits | Source |
|---|---|---|---|---|
| Voxtral TTS | Mar 2026 | Audio |
| Serenities AI |
| Mistral Large 3 | Dec 2025 | Flagship |
| IntuitionLabs |
| Voxtral | Jul 2025 | Audio |
| Mistral AI |
| Magistral | Jun 2025 | Reasoning |
| Mistral AI |
| Mistral Small 3 | Jan 2025 | Open release |
| Mistral AI |
| Ministral 3B & 8B | Oct 2024 | Edge |
| Mistral AI |
| Mistral Large 2 | Jul 2024 | Version update |
| Mistral AI |
| Codestral | May 2024 | Coding |
| Mistral AI |
| Mistral Large | Feb 2024 | Flagship |
| Mistral AI |
| Mixtral 8x7B | Dec 2023 | Open release |
| Mistral AI |
| Mistral 7B | Sep 2023 | Debut |
| Mistral AI |
Timeline
A practical timeline of model releases and company milestones.
This page tracks the moments that matter most: major model releases, version jumps and company moves that changed how this provider shows up in the AI market.
Voxtral TTS completes the audio loop with generation.
Mistral’s first text-to-speech model pairs with the Voxtral understanding line, rounding out a full voice stack.
Mistral Large 3 ships as the biggest open-weight MoE from a major lab.
Released with the dense Ministral 3 family, it doubles down on the bet that open frontier models can fund a business.
Voxtral opens a speech-understanding branch.
Open audio models give Mistral a voice story beyond text, feeding Le Chat’s real-time conversation features.
Magistral debuts Europe’s first frontier reasoning family.
Following Medium 3 and Devstral in May, the transparent multilingual reasoner completes a rapid first-half lineup refresh.
Mistral Small 3 returns the core line to Apache 2.0.
A 24B latency-optimized model re-commits the company to permissive licensing after a more commercial 2024.
Ministral 3B and 8B target edge and on-device use.
After Pixtral added vision in September, “les Ministraux” extend the portfolio to phones, robots and offline deployments.
Mistral Large 2 lands a week after Mistral NeMo.
A 123B dense flagship with 128K context closes much of the gap to GPT-4-class models in a single summer.
Codestral opens a dedicated coding branch.
Trained on 80+ languages with fill-in-the-middle support, it becomes the default open option for editor completion engines.
Mistral Large and Le Chat open the commercial chapter.
The flagship API model arrives with a consumer assistant and a Microsoft distribution deal, balancing the open-weight identity.
Mixtral 8x7B popularizes sparse mixture-of-experts.
The open MoE that outperforms GPT-3.5 makes “Mixtral” shorthand for efficient frontier-adjacent performance.
Mistral 7B launches via magnet link and beats Llama 2 13B.
A months-old startup ships the strongest small model in the world as a torrent, instantly defining European open-weight AI.