Mistral Large 4: "Le Chonk"
The preview opened on October 6 through Mistral Studio, and the model weights are due by the end of the month. The model uses a mixture-of-experts design with 1.05T total parameters, 52B of them active, plus a 1.6B vision encoder. Standard API pricing is $1.36 per million input tokens and $4.18 per million output tokens. OpenRouter currently lists it at half that. Mistral trained it from scratch on 3,800 NVIDIA Grace Blackwell GPUs in its own European datacenters. Artificial Analysis gave it 38 on its Intelligence Index, comparable to GPT-6 Luna (max, 38) and DeepSeek V4.1 Flash (max, 39). The same firm found it costs over 4x the cost per task of similar-intelligence open weights models. On the Vals Index it ranks #32 of 44 models. For founders, the draw is a near-frontier open model with a European deployment operated end to end by Mistral under European law. That matters when customers require data to stay in the EU. Because each task costs more, test it on your real workloads before switching.
