The Archive

Mistral Small 3: Apache 2.0, 81% MMLU, 150 tokens/s

Mistral AI releases Codestral 25.01, a code-generation model with API access for developers.

Mistral launches free multi-capability chat interface with search, vision, ideation, and coding features.

Mistral AI announces Pixtral Large, an upgraded multimodal model for vision and text tasks.

Mistral introduces Batch API enabling lower-cost asynchronous inference for builders.

Mistral AI releases moderation API for detecting undesirable text across policy dimensions.

Mistral announces edge-optimized small models, positioning them as best-in-class for edge deployment.

Mistral announces free API tier, reduced pricing, Mistral Small 3 enterprise, and free vision in le Chat.

Mistral releases Pixtral 12B, first multimodal Mistral model under Apache 2.0 license.

Mistral announces development tools and application-sharing platform for generative AI applications.

Mistral Large 2 flagship release with improved code, math, reasoning, multilingual support, and function calling.

Mistral AI releases Mistral NeMo, a 12B open-weights model with 128k context, built with NVIDIA under Apache 2.0.

Mistral AI releases Codestral Mamba, a Mamba2-based code generation model under Apache 2.0 license.

Mistral AI releases Mathstral, a 7B open-weights model specializing in mathematical reasoning and scientific tasks.

Mistral AI launches fine-tuning API and SDK enabling users to customize and deploy Mistral models.

Mistral AI announces virtual fine-tuning hackathon running June 5–30, 2024.

Mistral AI launches Codestral, a code generation model aimed at democratizing coding access.

Mistral AI introduces MNPL (Mistral Non-Production License) to balance openness in AI licensing.

Mistral AI publishes a general positioning statement on advancing accessible AI capabilities.

Mistral AI launches Le Chat, a conversational assistant in beta demonstrating its model capabilities.

Mistral AI positions Mistral Large as flagship reasoning model, now available on Azure.

Mistral AI launches inference API endpoints in early access for running models at scale.

Mistral AI releases Mixtral, a sparse mixture-of-experts model combining quality with efficient routing.

Mistral AI's mission statement on building open-weight models for frontier AI capabilities.

Mistral AI releases Mistral 7B under Apache 2.0, claiming best-in-class performance for the 7B parameter class.