Mistral AI Latest News
Mistral AI is a French open-model company focused on open weights and European sovereign AI.
-
-
OpenRouter benchmarks best OCR models: Claude Opus 5.5 most accurate, Mistral OCR best value
-
Mistral expands Mistral Studio to build enterprise agents, workflows and apps with Mistral Large 4 (Le Chonk)
-
Mistral Large 4 released with open weights and technical report
-
Unsloth releases Mistral Le Chonk 1-bit GGUF quantization, enabling local runs of a 1T-parameter model
-
Mistral Large 4 (Le Chonk) lands on Hugging Face: 1T-parameter MoE multimodal open-weight model
-
vLLM releases Day 0 support for Mistral Large 4
-
Mistral Large 4 enters Agent Arena top-15 labs, ranking 43rd
-
Mistral Large 4 enters Code Arena: WebDev at No. 45 with 1534 points
-
Mistral Large 4 enters Arena's Agent Arena and Code Arena evaluations
-
Mistral Large 4 launches public beta on OpenRouter: 1T parameters, 512K context
-
Mistral releases Mistral Large 4, Artificial Analysis calls it the smartest model outside the US and China
-
IMDEA Networks paper: nine conversational AI services leak conversation titles, prompts and screenshots to third parties
-
Mistral reviews methods and lessons from migrating 40,000 lines of Fortran 77 to C++ with AI agents
-
Mistral closes €3 billion Series D at over €21 billion valuation
-
Mistral releases Agentic Search, multi-step retrieval raises FinanceBench accuracy from 26.7% to 86%
-
Mistral open-sources 3B multimodal safety classifier Shieldstral, supporting custom policies at inference without retraining
-
Mistral AI releases Robostral Navigate embodied navigation model
-
PromptArmor discloses Microsoft Copilot Cowork Skill can bypass admin settings to call DeepSeek
-
Mistral releases Leanstral 1.5: a 6B activated-parameter formal proof model, fully open source under Apache-2.0
-
Artifacts 22: Zyphra, Cohere and Poolside are broadening the ecosystem
-
Weave launches intelligent model routing tool that plugs directly into Claude Code, Codex and Cursor
-
Mistral releases OCR 4 document understanding model with bounding boxes, block classification and confidence scores
-
OpenRouter: AI data residency compliance through API routing
-
Free LLM API Comparison: Rate Limits, Models and Real Costs (2026)
-
Hugging Face blog releases code-switching benchmark for speech agents
-
Mistral releases public preview of open-source Search Toolkit, unifying ingestion, retrieval, and evaluation in search pipelines
-
Mistral releases unified agent Vibe with Work and Code modes and a VS Code extension
-
Physical AI research that shapes industry
-
Mistral releases Mistral Medium 3.5, launches cloud remote coding agent and Work mode in Vibe and Le Chat
-
Mistral releases Connectors in Studio with built-in and custom MCP connectors
-
Mistral AI releases Voxtral TTS speech generation model with open weights
-
Mistral releases open-source Small 4 model with hybrid reasoning and image understanding
-
Mistral releases Small 4, unifying reasoning, multimodal and coding in an Apache 2.0 open-source model
-
Mistral releases Voxtral Transcribe 2, opens Voxtral Realtime weights under Apache 2.0
-
Mistral releases Vibe 2.0 terminal coding agent, Devstral 2 moves to paid API
-
Mistral releases Mistral OCR 3 with 74% overall win rate, priced at $2 per 1,000 pages
-
Mistral AI releases Devstral 2 (123B) and Devstral Small 2 (24B) coding models and Vibe CLI
-
Mistral releases Mistral 3 family: Mistral Large 3 and Ministral 3 all open-source
-
Mistral AI raises €1.7 billion Series C at €11.7 billion valuation, led by ASML
-
Mistral launches 20+ MCP connectors and Memories for Le Chat
-
Cohere releases Command A Reasoning enterprise reasoning model
-
Mistral releases open-source speech understanding model Voxtral in 24B and 3B versions
-
Mistral AI releases Devstral Medium and upgrades Devstral Small 1.1
-
Mistral AI releases first reasoning model Magistral, including open-source Magistral Small
-
Mistral releases Agents API with built-in code execution, web search, and multi-agent orchestration
-
Mistral AI and All Hands AI jointly release open-source coding model Devstral
-
Mistral AI releases Mistral Medium 3
-
Mistral releases Mistral Small 3.1, open-sourced under Apache 2.0 with 128k context support
-
Mistral releases OCR API, surpassing Google Document AI and Azure OCR on document understanding benchmarks
-
Mistral AI launches new le Chat on iOS and Android with Pro, Team, and Enterprise versions
-
Mistral releases 24B open-source model Mistral Small 3, matching Llama 3.3 70B performance under Apache 2.0
-
Mistral AI releases Codestral 25.01 with about 2x faster code generation
-
Mistral AI introduces web search, Canvas, document understanding, image generation, and Agent features for le Chat
-
Mistral releases open-weight multimodal model Pixtral Large
-
Mistral AI releases Ministral 3B and Ministral 8B on-device models
-
Mistral releases Mistral Small v24.09, launches free tier on la Plateforme and cuts prices across model lineup
-
Mistral releases open-source multimodal model Pixtral 12B under Apache 2.0 license
-
Mistral releases Mistral Large 2 with 123B parameters, 128k context, and open research-license weights
-
Mistral and NVIDIA jointly release 12B open-source model Mistral NeMo
Monthly archive
November 20264
October 20268
September 20263
August 20262
July 20263
June 20266
May 20263
April 20262
March 20263
February 20261
January 20261
December 20253
September 20252
August 20251
July 20252
June 20251
May 20253
March 20252
February 20251
January 20252
November 20242
October 20241
September 20242
July 20243
June 20241
May 20242
April 20241
February 20242
December 20233
September 20232
Frequently asked questions
- What is Mistral AI?
- Mistral AI is a French open-model company focused on open weights and European sovereign AI.
- What is the latest Mistral AI news?
- The most recent story is “How much can a looping AI agent burn? One test ran 41 times: 4 guards to set” (November 12, 2026). This page lists Mistral AI coverage in reverse chronological order so you can keep following it.
- How can I follow Mistral AI news?
- Open this page for the latest Mistral AI updates (72 stories). You can also subscribe to the site-wide RSS or JSON Feed; pages refresh about every 30 minutes.