Mistral Large 4 evaluated by Artificial Analysis with 512k context and API pricing

Artificial Analysis reported that Mistral Large 4 Preview scored 38 on its Intelligence Index, comparable to GPT-6 Luna at 38 and DeepSeek V4.1 Flash at 39. On cybersecurity benchmarks, the model registered 50% on the Artificial Analysis Cyber Index—matching GLM-5.3-Flash and trailing MiMo-V2.6-Pro at 56—while scoring 82% on CyberGym-E2E-AA. On document and image reasoning, it scored 19% on GDP.pdf.

The model provides a 512k-token context window, supports multimodal text and image inputs with text outputs, and accepts up to 100 images per request on Mistral's API, an increase from 8 in prior Mistral models. Standard API pricing is $1.36 per 1 million input tokens, $4.18 per 1 million output tokens, and $0.14 per 1 million cached input tokens. A 50% launch discount lowers rates to $0.68 and $2.09 per 1 million tokens for the first two weeks, reducing the calculated Cost per Task on the Intelligence Index to $0.57.

Mistral AI is serving the model in Research Public Preview on its API, with an open weights release for the 1-trillion parameter model (49 billion active parameters) scheduled for the end of October.