Mistral Small

Devstral-Small-2505 is a 24B parameter agentic LLM fine-tuned from Mistral-Small-3.1, jointly developed by Mistral AI and All Hands AI for advanced software engineering tasks. It is optimized for codebase exploration, multi-file editing, and integration into coding agents, achieving state-of-the-art results on SWE-Bench Verified (46.8%).

Devstral supports a 128k context window and uses a custom Tekken tokenizer. It is text-only, with the vision encoder removed, and is suitable for local deployment on high-end consumer hardware (e.g., RTX 4090, 32GB RAM Macs). Devstral is best used in agentic workflows via the OpenHands scaffold and is compatible with inference frameworks like vLLM, Transformers, and Ollama. It is released under the Apache 2.0 license.

Mistral Medium 3 is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances state-of-the-art reasoning and multimodal performance with 8× lower cost compared to traditional large models, making it suitable for scalable deployments across professional and industrial use cases.

The model excels in domains such as coding, STEM reasoning, and enterprise adaptation. It supports hybrid, on-prem, and in-VPC deployments and is optimized for integration into custom workflows. Mistral Medium 3 offers competitive accuracy relative to larger models like Claude Sonnet 3.5/3.7, Llama 4 Maverick, and Command R+, while maintaining broad compatibility across cloud environments.

Mistral Small 3.1 24B Instruct is an upgraded variant of Mistral Small 3 (2501), featuring 24 billion parameters with advanced multimodal capabilities. It provides state-of-the-art performance in text-based reasoning and vision tasks, including image analysis, programming, mathematical reasoning, and multilingual support across dozens of languages. Equipped with an extensive 128k token context window and optimized for efficient local inference, it supports use cases such as conversational agents, function calling, long-document comprehension, and privacy-sensitive deployments.

Mistral Saba is a 24B-parameter language model specifically designed for the Middle East and South Asia, delivering accurate and contextually relevant responses while maintaining efficient performance. Trained on curated regional datasets, it supports multiple Indian-origin languages—including Tamil and Malayalam—alongside Arabic. This makes it a versatile option for a range of regional and multilingual applications. Read more at the blog post here

Mistral Small 3 is a 24B-parameter language model optimized for low-latency performance across common AI tasks. Released under the Apache 2.0 license, it features both pre-trained and instruction-tuned versions designed for efficient local deployment.

The model achieves 81% accuracy on the MMLU benchmark and performs competitively with larger models like Llama 3.3 70B and Qwen 32B, while operating at three times the speed on equivalent hardware. Read the blog post about the model here.

Mistral's cutting-edge language model for coding. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation.

Learn more on their blog post: https://mistral.ai/news/codestral-2501/

Mistral Large 2 2411 is an update of Mistral Large 2 released together with Pixtral Large 2411

It provides a significant upgrade on the previous Mistral Large 24.07, with notable improvements in long context understanding, a new system prompt, and more accurate function calling.

This is Mistral AI's flagship model, Mistral Large 2 (version mistral-large-2407). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement here.

It supports dozens of languages including French, German, Spanish, Italian, Portuguese, Arabic, Hindi, Russian, Chinese, Japanese, and Korean, along with 80+ coding languages including Python, Java, C, C++, JavaScript, and Bash. Its long context window allows precise information recall from large documents.

Pixtral Large is a 124B parameter, open-weight, multimodal model built on top of Mistral Large 2. The model is able to understand documents, charts and natural images.

The model is available under the Mistral Research License (MRL) for research and educational use, and the Mistral Commercial License for experimentation, testing, and production for commercial purposes.

Ministral 3B is a 3B parameter model optimized for on-device and edge computing. It excels in knowledge, commonsense reasoning, and function-calling, outperforming larger models like Mistral 7B on most benchmarks. Supporting up to 128k context length, it’s ideal for orchestrating agentic workflows and specialist tasks with efficient inference.

Ministral 8B is an 8B parameter model featuring a unique interleaved sliding-window attention pattern for faster, memory-efficient inference. Designed for edge use cases, it supports up to 128k context length and excels in knowledge and reasoning tasks. It outperforms peers in the sub-10B category, making it perfect for low-latency, privacy-first applications.

The first multi-modal, text+image-to-text model from Mistral AI. Its weights were launched via torrent: https://x.com/mistralai/status/1833758285167722836.

A 7.3B parameter Mamba-based model designed for code and reasoning tasks.

A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA.

The model is multilingual, supporting English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, and Hindi.

It supports function calling and is released under the Apache 2.0 license.

A high-performing, industry-standard 7.3B parameter model, with optimizations for speed and context length.

Mistral 7B Instruct has multiple version variants, and this is intended to be the latest version.

A high-performing, industry-standard 7.3B parameter model, with optimizations for speed and context length.

An improved version of Mistral 7B Instruct v0.2, with the following changes:

NOTE: Support for function calling depends on the provider.

Mistral's official instruct fine-tuned version of Mixtral 8x22B. It uses 39B active parameters out of 141B, offering unparalleled cost efficiency for its size. Its strengths include:

See benchmarks on the launch announcement here. #moe

Mixtral 8x22B is a large-scale language model from Mistral AI. It consists of 8 experts, each 22 billion parameters, with each token using 2 experts at a time.

It was released via X.

#moe

This is Mistral AI's flagship model, Mistral Large 2 (version mistral-large-2407). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement here.

It supports dozens of languages including French, German, Spanish, Italian, Portuguese, Arabic, Hindi, Russian, Chinese, Japanese, and Korean, along with 80+ coding languages including Python, Java, C, C++, JavaScript, and Bash. Its long context window allows precise information recall from large documents.

This is Mistral AI's closed-source, medium-sided model. It's powered by a closed-source prototype and excels at reasoning, code, JSON, chat, and more. In benchmarks, it compares with many of the flagship models of other companies.

Note: This model is being deprecated. Recommended replacement is the newer Ministral 8B

This model is currently powered by Mistral-7B-v0.2, and incorporates a "better" fine-tuning than Mistral 7B, inspired by community work. It's best used for large batch processing tasks where cost is a significant factor but reasoning capabilities are not crucial.

A high-performing, industry-standard 7.3B parameter model, with optimizations for speed and context length.

An improved version of Mistral 7B Instruct, with the following changes:

Mixtral 8x7B Instruct is a pretrained generative Sparse Mixture of Experts, by Mistral AI, for chat and instruction use. Incorporates 8 experts (feed-forward networks) for a total of 47 billion parameters.

Instruct model fine-tuned by Mistral. #moe

A 7.3B parameter model that outperforms Llama 2 13B on all benchmarks, with optimizations for speed and context length.