GoKawiilMistral Large 4 ranks top outside US/China, lags Chinese open models overall
Mistral released Large 4, a trillion-parameter mixture-of-experts model, in a research preview this week, with full weights due at the end of October. Independent benchmarking firm Artificial Analysis scored it 38 on its Intelligence Index, making it the top model from outside the US and China, though several Chinese open models including DeepSeek V4.1 Flash and MiMo-V2.6-Pro scored as high or higher overall. Large 4 scored 50 on Artificial Analysis's Cyber Index, tying GLM-5.3-Flash, and hit 82% on the CyberGym-E2E-AA cybersecurity test, its best individual result.
GoKawiil's interpretation of the reporting above, not reported fact.
The results suggest Mistral remains competitive on cybersecurity-related tasks even as Chinese labs continue to close or exceed the gap on general intelligence benchmarks, reinforcing a narrative that frontier open-weight AI development is increasingly concentrated among US and Chinese firms. Artificial Analysis also noted the model's cost per task is over four times higher than similarly capable open models, which could affect its appeal to developers weighing performance against price once the weights are public.
- Mistral Large 4 scored 38 on Artificial Analysis's Intelligence Index, the highest among non-US, non-Chinese models.
- Several Chinese open models, including DeepSeek V4.1 Flash and MiMo-V2.6-Pro, matched or outperformed Large 4 overall.
- Large 4's weights are expected to be released publicly at the end of October, following this research preview.
Source: tomshardware.com — Shane Downing, 2026-10-09
Published there as: “Mistral’s new Large 4 trails some Chinese open models in independent tests”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.