Models
Mistral text-to-speech model with zero-shot voice cloning and multilingual support.
Mistral Codestral Embed model for code embeddings and retrieval.
State-of-the-art semantic model for extracting representation of text extracts.
Mistral code model for fill-in-the-middle, code generation, and coding assistants.
Mistral code-agent model for software engineering tasks and tool-using coding agents.
Thinking model excelling in domain-specific, transparent, and multilingual reasoning.
Thinking model excelling in domain-specific, transparent, and multilingual reasoning.
Mistral Ministral compact model for edge and on-device deployment.
Mistral Ministral 3 compact open model for edge and local deployment.
Mistral largest Ministral 3 open model with strong local-deployment performance.
Mistral compact open model for edge and local deployment.
Mistral efficient open model with strong text and vision capabilities.
Mistral Large high-capability model for reasoning and complex enterprise workloads.
Mistral Large 2 is a high-capability Mistral model for reasoning and complex enterprise workloads.
Mistral open-weight general-purpose multimodal model with a sparse MoE architecture.
Mistral frontier-class multimodal model optimized for agentic and coding use cases.
Mistral Medium model (latest alias) balancing quality and cost for enterprise workloads.
Mistral Nemo is a 12B open model designed for multilingual and efficient text generation workloads.
Mistral Small is a cost-efficient Mistral model for fast, high-volume workloads.
Mistral hybrid model unifying instruct, reasoning, and coding capabilities.
Mistral OCR service for extracting interleaved text and images from documents.