Mistral Large 2 was released July 24, 2024 as a 123B parameter model. Scored 84% on MMLU and 92% on HumanEval. Competitive with GPT-4o and Claude 3.5 Sonnet at significantly lower cost. Featured a 128K context window and was released under the Mistral Research License (later made more permissive). Available via API and self-hosting.
Pricing
| Type | Price per MTok |
|---|---|
| Input | $2 |
| Output | $6 |
Benchmarks6
| Benchmark | Score |
|---|---|
| MMLU | 84% |
| HumanEval | 92% |
| MATH | 69.1% |
| GPQA Diamond | 43.9% |
| IFEval | 80.1% |
| BBH | 78.8% |
Mistral Family1
| Model | Tier | Released | Input $/MTok |
|---|---|---|---|
| Mixtral 8x7B | — | 2023-12-11 | — |
Details
Model FamilyMistral
Tierlarge
Generation2
Release Date2024-07-24
Parameters123B
Context Window128K tokens
Open WeightYes
Modalitytext
Capabilities1
tool-use
Sources1
Tags
mistralfrontier-aiopen-weight