LLM APIEU-hostedFree tier
Mistral Large · Mistral AI
Quick answer
Mistral Large is EU-hosted frontier model from Paris — the go-to option for GDPR-sensitive workloads. It is made by Mistral AI, priced as “pay per token (from $3 / 1m input tokens)”, with observed latency around 350 ms, and offers EU data residency. Rated 4.4/5 by our editorial review.
Overview
Mistral Large is a frontier-class model built and hosted in the EU, making it one of the few credible options for organizations with strict EU data-residency requirements. Strong multilingual support across European languages.
Benchmarks
| MMLU (5-shot) | 84.0% |
|---|---|
| EU data residency | Yes (EU region) |
| Median latency (observed) | ~350 ms TTFT |
| Context window | 128K tokens |
Pros
- EU-hosted with EU data residency options
- Strong multilingual (European) performance
- Competitive token pricing
Cons
- Slightly behind GPT-4o / Claude on some benchmarks
- Smaller ecosystem and tooling