Mistral Large 4 (Le Chonk) Becomes Top Non-Chinese Open Model On Artificial Analysis Intelligence Index

Mistral Large 4 Preview, nicknamed "Le Chonk", scored 38 on the Artificial Analysis Intelligence Index v4.3.2, making it the highest-scoring open-weights model from outside China on the chart, tied with OpenAI's GPT-6 Luna. The 1-trillion-parameter mixture-of-experts model with 49 billion active parameters trails Chinese open models including Xiaomi's MiMo-V2.6-Pro at 46, Zhipu's GLM-5.3 at 45, Moonshot's Kimi K3 at 44 and DeepSeek V4.1 Flash at 39, while leading Western open models such as Thinking Machines' Inkling at 25 and Nvidia's Nemotron 3 Ultra at 23. Mistral Large 4 is available through Mistral's API now, with open weights due at the end of October.

After a few months in the wilderness, Mistral appears to be leading the Western open model pack once again. Mistral’s newest flagship has landed on the Artificial Analysis Intelligence Index, and it’s enough to make it the highest-scoring open model from outside China on the chart. Mistral Large 4 https://officechai.com/ai/mistral-large-4-big-chonk/ Preview, nicknamed “Le Chonk”, scores 38 on the index, putting it ahead of the other Western open-weights models on the leaderboard, though still well short of the best Chinese open models. Where Large 4 lands On the latest version of the index v4.3.2, which combines 10 evaluations , Large 4 Preview sits 18th among the 25 models shown on Artificial Analysis’s chart, tied with OpenAI’s GPT-6 Luna at 38. At the top, Claude Opus 5.5 leads with 58 https://officechai.com/ai/claude-opus-5-5-creates-5-point-lead-over-gpt-6-astra-jumps-to-top-spot-on-artificial-analysis-intelligence-index/ , followed by a cluster at 53 that includes Claude Fable 5.1, GPT-6 Astra and Gemini 4 Argon. Among Western open-weights models, Mistral Large 4 has a clear lead. Nvidia’s Nemotron 3 Ultra scores 23 on the same chart, and Thinking Machines’ Inkling, which debuted as the top US open model https://officechai.com/ai/thinking-machines-inkling-becomes-top-us-open-model-on-artificial-analysis-intelligence-index/ on an earlier version of the index, is at 25. Because Artificial Analysis has been revising the index frequently, scores from different versions aren’t directly comparable, which is why Inkling’s number is lower here than at its launch. Still behind China’s best The Chinese open-weights field remains ahead. Xiaomi’s MiMo-V2.6-Pro, which took the top open-model spot https://officechai.com/ai/xiaomi-mimo-v-2-6-pro-benchmarks/ last month, scores 46, eight points above Large 4. Zhipu’s GLM-5.3 is at 45, Moonshot’s Kimi K3 at 44, and the smaller GLM-5.3-Flash at 42. Even DeepSeek V4.1 Flash edges past Mistral with a 39. That puts Large 4 in a familiar position. The gap between Chinese and Western open models has been a recurring theme on the index since Kimi K2.5 pulled ahead of the field https://officechai.com/ai/kimi-k2-becomes-top-open-model-on-artificial-analysis-index/ at the start of the year, and Large 4 narrows that gap without closing it. Mistral’s own claim, made at launch, was that Large 4 is the best open-weights model from the US or Europe on aggregated benchmarks, and the Artificial Analysis result is consistent with that framing. What Mistral’s model offers Large 4 is a 1-trillion-parameter mixture-of-experts model with 49 billion active parameters, and it is natively multimodal. Mistral says it’s state-of-the-art on workloads like cyber defense, manufacturing and finance, and that it surpasses closed frontier models on visual grounding. The model is available through Mistral’s API now and can be deployed in Europe on Mistral’s own cloud infrastructure. The open weights are due at the end of October. The release comes as open models take a growing share of real-world usage. Open-weight models now handle most token volume on Vercel’s AI Gateway https://officechai.com/ai/share-of-closed-models-has-fallen-from-around-70-to-21-in-the-last-3-months-vercel-data/ , so a strong open option from a European lab, one that can be self-hosted once the weights ship, gives enterprises that can’t or won’t use Chinese models a more capable alternative. Mistral, which raised $2 billion in a round led by ASML https://officechai.com/ai/mistral-raises-2-billion-in-new-funding-round-led-by-asml/ , has made that sovereignty argument central to its pitch.