Mistral API pricing in 2026

Mistral AI offers strong models at prices often lower than OpenAI or Anthropic. Mistral Large targets demanding tasks (reasoning, long output); “Mistral Small” and open-source variants target cost and latency. The API is billed per million tokens (input and output), sometimes with volume discounts or plans.

Strengths: good quality/price on many benchmarks, multilingual models, long context. Use cases: writing, code, analysis, chat. To choose, compare Mistral’s pricing with OpenAI, Anthropic, and Google by volume and request type (short vs long, input-heavy vs output-heavy). An up-to-date comparator of price per million tokens helps you decide.