Mistral Large 4

The Mistral Large 4 family from Mistral AI introduces natively multimodal Mixture-of-Experts model that unify instruction following, reasoning, coding, and agentic workflows. It combines configurable reasoning effort, million-token context support, and fluency in more than 160 languages for enterprise and security workloads.

1 variant

Mistral Large 4 1T A52B

Mistral

Params: 1T

Mistral Large 4 1T A52B is a natively multimodal Mixture-of-Experts model that unifies instruction following, reasoning, coding, and agentic workflows. It holds approximately 1T total parameters with 52B active, using a 72-layer architecture with 6,144 hidden size, 64 attention heads, and 64 KV heads. Sparse indexed attention runs in all layers, with rotary embeddings omitted every fourth layer, and each MoE layer activates 4 of 128 routed experts plus a shared expert. Supporting a 1M-token context, it adds a Pixtral-style vision encoder and an embedded DSpark draft head.

Text GenerationInstruction Following+8

Runs onB200+

Runs on

Mistral Large 4 Models | Vultr Docs