LLMOpen weights
Mistral Large 2
Mistral AI
Updated May 21, 2026 · Benchmarks via Artificial Analysis, specs & pricing via OpenRouter · Methodology · Report an error
Superseded by Mistral Large 3
Specifications
- Type
- LLM
- Access
- Open weights
- Released
- July 24, 2024
- License
- Mistral Research License
- Parameters
- 123B
- Context window
- 128,000 tokens
- Output speed
- 42 tok/s
- Latency (TTFT)
- 0.4s
- Input
- Text
- Output
- Text
- API pricing
- $2.00 in · $6.00 out / 1M tokens
Capabilities
Function callingStructured output
Benchmarks
Our take
Mistral’s flagship dense model, strong on multilingual and code tasks, with weights available under a research license — a European counterweight in the open-model field.