01 / The short version
What happened
Mistral Large 4, also known as Le Chonk, is a multimodal, multilingual MoE model from Mistral that accepts text and images to generate text. It features approximately 1T total parameters, 49B active parameters, and a 1M token context window. The model uses a Pixtral vision encoder and sparse attention. Mistral reports an 82% score on the Artificial Analysis Cyber Index and coding results including 61.7% on DeepSWE v1.1 and 59.4% on SWE-Atlas-QnA. It is distributed under the Mistral Ecosystem Open-Weight License and supports transformers, vLLM, and llama.cpp.
See the exact references02 / Key takeaways
What you need to know
- 01
Mistral Large 4 has about 1T total parameters and 49B active parameters.
- 02
Mistral reports an 82% score on the Artificial Analysis Cyber Index.
- 03
The model supports text and image inputs with a 1M token context window.
Keep in perspective
What to watch for
Performance claims are attributed to Mistral and have not been independently validated.
Go to the source
Exact references
These are the original pages used for this brief. Publisher claims are not independent evaluations.
01Primary source · Hugging FaceRead the original announcementhttps://huggingface.co/blog/mistral-large4AI-generated from the linked source. It can miss context; verify consequential details in the original. How the radar works