Mistral Large
by Mistral AI · Model · Language & code models
weak
Trails rival models on intelligence and costs more, per independent tests
Mistral Large 4: 1T-parameter MoE with 52B active, available by API; weights promised at end of month
Why we say so
Artificial Analysis scores Mistral Large 4 below Kimi K3, GLM 5.3 and DeepSeek Flash on its Intelligence Index, and says it costs more. It also has worse intelligence-versus-cost than MiMo 2.6 Pro and DeepSeek 4.1 Flash. Mistral's own claim is a 1T-parameter model with 52B active, available by API, with weights promised but not yet on record.