Mistral Large 4 scores 38 on Artificial Analysis, behind four Chinese open models

First reported by Tom's Hardware at · Updated · 2 sources

Artificial Analysis gave Mistral's Large 4 a score of 38, according to Tom's Hardware, placing it behind open models from Xiaomi, Z.ai, Moonshot and DeepSeek. Separately, Sifted reports that Mistral and other developers face a growing risk of open-model 'abliteration'. The ranking comes from independent testing of the new model.

Covered by 2 publishers within 5 hours of the first report.

Reporting2

Tom's Hardware Mistral’s new Large 4 trails some Chinese open models in independent tests · Shane Downing
Sifted Report: Mistral and others increasingly at risk of open model ‘abliteration’