Mistral AI API
Efficient open-source LLMs from Europe, Mistral & Mixtral
Freemium
✓ Verified
🇪🇺 European
Mistral AI is a French startup that has rapidly become Europe’s leading LLM provider. Their models — including Mistral Large and the Mixtral Mixture-of-Experts series — offer competitive performance to GPT-4 class models at significantly lower pricing. Pricing starts at $0.27 per million input tokens for Mistral Small. Available via API, AWS Bedrock, Azure AI, and GCP. Mistral is the preferred choice for European enterprises requiring GDPR compliance and EU data residency. Open-weight models are also available for self-hosting.
API Details
Auth Method
API Key
Pricing Model
Freemium
Free Tier
Yes — free tier on La Plateforme
Rate Limit
5 RPM (free), up to 2000 RPM (paid)
Format
REST / JSON (OpenAI-compatible)
Versioning
Model-based (mistral-large-latest, mixtral-8x7b)
SLA / Uptime
99.5%
Compliance
GDPR (EU data residency available)
Geographic Restrictions
Global; EU data residency option for enterprise
Last Verified
2026-02-20
Categories
Frequently Asked Questions
Mistral offers a tiered model range: Mistral 7B (open-source, free to self-host), Mistral Small (fast and affordable), Mistral 8x7B/8x22B (mixture-of-experts), Mistral Large (flagship), and Codestral (code specialist). All are available via the La Plateforme API, with open-source models also on Hugging Face.
Yes u2014 Mistral is a French company and all data is processed in the EU, which makes it attractive for GDPR-sensitive applications. Mistral Large 2 competes with GPT-4o on benchmarks while keeping data within EU jurisdiction. This is a significant advantage for healthcare, finance, and legal applications in Europe.
Mistral is generally cheaper than OpenAI. Mistral Small costs $0.20/$0.60 per million tokens. Mistral Large costs $2/$6 per million tokens, compared to GPT-4o's $2.50/$10. Codestral is u20ac1/$1/$3 per million tokens for code generation tasks u2014 a strong value proposition for developer tooling.
Yes. Mistral 7B, Mixtral 8x7B, and Mixtral 8x22B are fully open-source under the Apache 2.0 licence and can be run locally using Ollama, llama.cpp, or vLLM. This makes Mistral ideal for applications requiring full data privacy, offline use, or custom fine-tuning without API costs.