Overview
Mistral Large 2 is the second generation of the flagship model from Paris-based Mistral AI, released in July 2024. With 123 billion parameters, it was designed to run on a single high-end server node and emphasized coding, reasoning, function calling and dozens of programming and human languages.
Highlights
- 123 billion parameters
- 128K-token context window
- Support for many human languages and over 80 programming languages
- Native function calling
- Weights available under a research license; commercial use via Mistral’s platform or a commercial license
Typical uses
Enterprise assistants, coding help, multilingual applications and agent workflows, particularly for organizations preferring a European provider.
Things to consider
- The research license does not allow commercial self-hosting without a separate agreement.
- Mistral has released newer models since; check the current lineup.
- Self-hosting needs substantial GPU memory.
Specifications are taken from the provider’s official announcement, documentation or model card. Benchmark results depend on evaluation settings; see how AI benchmarks work.