← Back to live feed1 story
1
Mistral Says Its New AI Model ‘Le Chonk’ Is the Best Open-Weight Offering Outside of China
wired.comMistral released a new trillion-parameter model called Le Chonk, aiming to demonstrate it remains competitive in building frontier-level artificial intelligence.
Sign in to suggest edits
Key sources
- SOURCE@mistralai“1T parameters, natively multimodal. 49B active.”x.com
- SUPPORT@guillaumelample“trained on 3,800 NVIDIA Grace Blackwell GPUs in our European datacenters”x.com
- SUPPORT@artificialanlys“most intelligent model from outside the US and China, ahead of countries such as South Korea and the United Arab Emirates”x.com
- SOURCE@openrouter“1T params (49B active), natively multimodal, 512K context, up to 256K output”x.com
- SUPPORT@petergostev“Mistral Large 4 was trained on 4,000 GPUs and Astra was trained on 100,000 of them”x.com
- SUPPORT@hesamation“trained on 3,800 NVIDIA Grace Blackwell GPUs in European data centers”x.com
- SUPPORT@nielsrogge“no blog post, no benchmarks”x.com
- SUPPORT@qtnx_“the RL run is still continuing and we keep seeing improvements”x.com