Mistral Large 4 Ranks Eighth Among Open Models, Trails Seven Chinese Rivals

French startup Mistral AI released its Large 4 model on October 6, claiming it pushes the frontier of open-weight performance. Independent benchmarking service Artificial Analysis scored the model 38.4 points on its Intelligence Index, making it the highest-ranked open model from outside the United States and China.

However, seven Chinese open-weight models rank ahead of Large 4. Xiaomi's MiMo-V2.6-Pro leads with 46.3 points, followed by Z.ai's GLM-5.3 at 44.8 and Moonshot AI's Kimi K3 at 43.6. The previous non-Chinese open leader, South Korea's Motif 3, scored 33.6 points.

“Mistral Large 4 is the most intelligent model from outside the US and China.”

Mistral Large 4 uses a mixture-of-experts architecture with 1 trillion total parameters and 49 billion active parameters, trained on Nvidia Grace Blackwell GPUs. The model weights are scheduled for release at the end of October; until then, Artificial Analysis lists the preview as a proprietary model.

In cybersecurity testing, Large 4 scored 50 on the Artificial Analysis Cyber Index, tying Z.ai's GLM-5.3-Flash and trailing MiMo-V2.6-Pro's 56. It achieved 82% on the CyberGym-E2E-AA end-to-end measure, surpassing MiMo-V2.6-Pro's 79% and GPT-6 Luna's 78%.

The model carries a standard API price of $1.36 per million input tokens and $4.18 per million output tokens. Artificial Analysis calculates a cost of $1.13 per Intelligence Index task, which it notes is over four times higher than similar-intelligence open-weight models.

Closed proprietary models maintain a significant lead. Anthropic's Claude Opus 5.5 scores 57.6, OpenAI's GPT-6 Astra 52.7, and Google's Gemini 4 Argon 52.6 on the Intelligence Index, leaving Large 4 more than 19 points behind the world's best systems.