Mistral Small 3.1 Release Notes

Mistral AI has released Mistral Small 3.1, a high-performance small-weight model that integrates multimodal understanding and an expanded 128k token context window. This release is significant because it is the first open-source model to surpass leading small proprietary models in text performance, multimodal capabilities, and latency across multiple dimensions.

Technical Specifications and Performance

Mistral Small 3.1 outperforms comparable models such as GPT-4o Mini and Gemma 3, while maintaining high inference speeds of 150 tokens per second. The model is designed to balance low latency and cost efficiency with the ability to handle text, multimodal inputs, and multiple languages.

Hardware Requirements

Mistral Small 3.1 is optimized for lightweight deployment. It can run on a single NVIDIA RTX 4090 GPU or a Mac with 32GB of RAM, making it highly suitable for on-device AI applications.

Context and Licensing

  • Context Window: The model supports up to 128k tokens, allowing for the processing of longer documents and more complex prompts.
  • Licensing: Mistral Small 3.1 is released under the Apache 2.0 license, providing broad accessibility for open-source development and commercial use.

Core Capabilities and Use Cases

Mistral Small 3.1 is a versatile model capable of instruction following, conversational assistance, image understanding, and function calling. It serves as a foundation for both consumer-grade and consumer-grade AI applications.

Key Application Areas

  • Conversational AI: Fast-response virtual assistants where quick and accurate responses are critical.
  • Agentic Workflows: Low-latency function calling for rapid execution within automated systems.
  • Domain Specialization: The model can be fine-tuned for specialized fields such as medical diagnostics, legal advice, and technical support.
  • Multimodal Enterprise Tasks: The model supports image-based customer support, document verification, diagnostics, visual inspection for quality checks, and object detection in security systems.

Model Availability and Ecosystem

Mistral AI has released both the pretrained base model and the instruct checkpoint to enable downstream customization and the development of advanced reasoning models.

Distribution Channels

  • Open Weights: Available for download on Hugging Face (Mistral Small 3.1 Base and Mistral Small 3.1 Instruct).
  • API Access: Available via the developer playground, La Plateforme.
  • Cloud Platforms: Available on Google Cloud Vertex AI, with upcoming availability on NVIDIA NIM and Microsoft Azure AI Foundry.

Sources

Related

  • Dispatch
    Mistral Small 3 Release Notes

    Mistral AI has released Mistral Small 3, a latency-optimized 24B-parameter model under the Apache 2.0 license that rivals the performance of models three times its size.

  • Dispatch
    Mistral Small 4 release notes / what's new

    Mistral AI has released Mistral Small 4, a unified open-source model under Apache 2.0 that integrates reasoning, multimodal, and agentic coding capabilities into a single versatile architecture.

  • Dispatch
    Mistral AI Devstral Models Release

    Mistral AI has released Devstral Medium and Devstral Small 1.1, providing state-of-the-art agentic coding capabilities with a focus on generalization across different prompts and agentic scaffolds.

  • Dispatch
    Mistral Large and Mistral Small Release

    Mistral AI has released Mistral Large, a flagship model with top-tier reasoning and multilingual capabilities, and Mistral Small, a model optimized for low-latency workloads.

  • Dispatch
    Mistral AI Voxtral Release

    Mistral AI has released Voxtral, a family of open-weight speech understanding models in 24B and 3B sizes that provide state-of-the-art transcription and native semantic understanding under the Apache 2.0 license.