Mistral
Mistral is the leading European AI company, founded in 2023 by former Google DeepMind and Meta researchers and based in Paris, and its lineup has moved almost entirely to open weights. Mistral Medium 3.5, released April 28, 2026 at $1.50 input and $7.50 output per 1M tokens with a 256K context window, is the model Mistral calls frontier-class for agentic and coding work. Mistral Large 3 is a 675 billion parameter open-weight mixture-of-experts with 41 billion active under Apache 2.0 at $0.50 and $1.50, and Mistral Small 4 folds instruct, reasoning, and coding into a 119 billion parameter model with 6.5 billion active at $0.15 and $0.60, also Apache 2.0. Retirements have been steady: Large 2.1 on May 31, Small 3.2 on July 31, and Medium 3 and 3.1 on August 31, 2026. On September 8, 2026 Mistral raised €3 billion in a Series D at a post-money valuation above €21 billion, which it calls the largest equity round ever completed by a European technology company. Samsung Electronics led, with Scaleup Europe Fund (managed by EQT) and PSG Equity as co-leads and NVIDIA and ASML among the existing investors that participated. The pitch has shifted with the money: European enterprises that need data sovereignty still buy Mistral, but the company now sells the full stack, from open weights to its own compute, as insurance against single-vendor lock-in.
Service status
Loading…
Founded
2023
Headquarters
Paris, France
CEO
Arthur Mensch
Models
3 active
Key Products
Strengths
- ✓European data sovereignty
- ✓Apache 2.0 open weights on Large 3 and Small 4
- ✓EUR 3B Series D led by Samsung Electronics
- ✓Competitive pricing
- ✓Growing enterprise presence
Mistral Models
| Model | Input / 1M | Output / 1M | Context | Capabilities |
|---|---|---|---|---|
| Mistral Medium 3.5 | 1.50 | 7.50 | 256K | text, code, tool-use, reasoning |
| Mistral Large | 0.50 | 1.50 | 262K | text, vision, tool-use, code |
| Mistral Small | 0.15 | 0.60 | 262K | text, tool-use, code |
Prices per 1M tokens in USD. See the full pricing guide for detailed analysis.
Benchmark Scores
| Model | SWE-bench | MMLU-Pro | HumanEval | GPQA Diamond | MATH | OSWorld 2.0 | BrowseComp | FrontierCode v1.1 | Terminal-Bench 4.0 | Humanity's Last Exam (tools) |
|---|---|---|---|---|---|---|---|---|---|---|
| Mistral Medium 3.5 | 77.6 | N/A | N/A | N/A | N/A | N/A | N/A | 8.0 | N/A | N/A |
| Mistral Large | 47.2 | 86.8 | 89.1 | 57.3 | 80.4 | N/A | N/A | N/A | N/A | N/A |
| Mistral Small | N/A | 78.4 | 82.5 | 44.6 | 68.9 | N/A | N/A | N/A | N/A | N/A |
Comparisons
Is Mistral Down?
Real-time status monitoring for Mistral services, updated every 2 minutes.