Mistral Pricing Calculator (Large, Medium, Small, Codestral)

Every current Mistral API model and its price per million tokens. Mistral is the cheapest EU-hosted frontier provider and is roughly 5–50× cheaper than OpenAI's flagship tiers on a like-for-like basis. For monthly bill projections plug your traffic into the full cost calculator or compare Mistral side-by-side with other providers on the comparison tool.
ModelInput / 1MOutput / 1MContext
Mistral Medium 3.5
Mistral
$0.40$2.00256K
Mistral Large
Mistral
$2.00$6.00128K
Mistral Small 24B
Mistral
$0.05$0.0833K
Codestral 2508
Mistral
$0.30$0.90256K

Which Mistral model should you use?

Mistral Large — general-purpose reasoning, RAG, agentic tool use. The default if you need GPT-4o-class quality at half the price. Mistral Medium 3.5 — best price/intelligence ratio in the Mistral lineup; switch here from Large when you don't need maximum reasoning depth. Mistral Small 24B — high-volume classification, extraction, structured outputs at 50× the cost-efficiency of frontier tiers. Codestral 2508 — purpose-built for code, with FIM (fill-in-middle) support for IDE integrations.

EU data residency advantage

Mistral's API runs entirely on EU infrastructure, which makes it the only viable frontier choice for many German, French and BaFin-regulated workloads. No US Cloud Act exposure, no Schrems II issues. For US teams that don't need data residency, OpenAI and Gemini are price-competitive on the cheap tiers.

Related guides

Compare with Claude pricing, see the cheapest LLM ranking, or read LLM Price Comparison 2026.

Frequently asked questions