Mistral launches Mistral Large 4 preview, a 1T-parameter open-weight model
Read original source →mistral.ai
Mistral Large 4 ("Le Chonk") is a 1T-parameter natively multimodal model with 49B active parameters, trained on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's European datacenters. The preview API is live on Mistral Studio; weights drop end of this month.
How this story unfolded
1 day · 13 reports · 27 community posts · 40 of 46 shown
- Oct 6
Introducing Mistral Large 4mistral.ai
Mistral Says Its New AI Model ‘Le Chonk’ Is the Best Open-Weight Offering Outside of Chinawired.com
Mistral debuts Large 4 ‘Le Chonk', a 1-trillion parameter text output model with high benchmarks planned for open weights releaseventurebeat.com
Mistral unveils new AI model it says rivals best open systems from Chinacnbc.com
Mistral’s new 1T model aims to leapfrog closed and open rivalstechcrunch.com
Mistral Large 4 now available on AI Gatewayvercel.com
Mistral AI Releases Mistral Large 4 (Le Chonk): A 1.05T Parameter Multimodal MoE Modelmarktechpost.com
Mistral’s new AI tried to escape its test environment. In three weeks, anyone can download itthenewstack.io
Mistral AI Drops 'Le Chonk': A Massive AI Model Named After a Cat Memedecrypt.co
Mistral Large 4simonwillison.net
- Oct 7
More stories today
Anthropic details Claude dynamic workflows at AI Engineer NYC
Anthropic's CJ Avilla and Harrison Stall presented dynamic workflows, a new multiagent orchestration feature announced Friday, at AI Engineer New York 2026. Flows let Claude choose the organization of agents for workloads longer than a single agent or roster.
Latent Space·2 hours ago
Essay urges developers to avoid OpenAI and Anthropic models
Security engineer Micah Lee argues open-weight models now match the latest ChatGPT and Claude on complex tasks like vulnerability research, citing private hosting via secure enclaves (Tinfoil) and provider switching as advantages.
Hacker News·2 hours agoDeveloper builds receipt-printing mystery game on M5StickS3 with Codex
dani bauer·2 hours ago
Perplexity Projects implements Git-based filesystem for agent collaboration
Perplexity·3 hours ago
Microsoft and Musk race to reinvent the AI chip
Bloomberg's Ed Ludlow breaks down Microsoft's latest homegrown Maia chips, aimed at reducing reliance on outside semiconductors for AI. Elon Musk is testing as many as ten new chip designs a year.
Bloomberg Technology·3 hours ago
OpenAI reopens Codex Ambassador applications
OpenAI Developers·3 hours ago
OpenAI GPT-6 Sol and Luna reach Bedrock via cross-Region inference
Amazon Bedrock now serves OpenAI's GPT-6 Sol and GPT-6 Luna through global cross-Region inference for teams in Thailand, Malaysia, Singapore, Indonesia, and Taiwan. The setup targets throttling that occurs when single-Region inference demand exceeds local capacity.
AWS AI Blog·3 hours ago
FrontierCode 1.1 leaderboard launches for Cognition's coding eval
Artificial Analysis·3 hours ago
