Mistral Large 2
Open SourceMistral AI · 2024-07
Mistral AI's flagship open-weight model with 123B parameters and multilingual excellence.
Quick Facts
Parameters
123B
Context Window
128K tokens
Modalities
text
Open Source
Yes
License
Mistral Research License (open-weight)
Pricing
Free (open-weight) / API from $2.00/1M tokens
Released
2024-07
Developer
Mistral AI
About
Mistral Large 2 is Mistral AI's flagship open-weight model with 123 billion parameters. It offers strong performance across reasoning, coding, and multilingual tasks with support for French, German, Spanish, Italian, Portuguese, Arabic, Hindi, Russian, Chinese, Japanese, and Korean. Built with a focus on efficiency, it features a 128K token context window and supports function calling and JSON output. Mistral Large 2 is competitive with GPT-4o and Claude 3.5 Sonnet on many benchmarks while being available under an open-weight license for self-hosting and customization.
Strengths
- +Exceptional multilingual performance across 12+ languages
- +Open-weight for self-hosting and customization
- +Strong coding and reasoning benchmarks
- +Efficient architecture for its parameter count
Weaknesses
- −Text-only model, no vision capabilities
- −Research license has some restrictions
- −Smaller ecosystem than Llama models
Best For
Multilingual AI applications across European languages
Self-hosted enterprise AI with strong reasoning
Function calling and structured output applications
Cost-effective alternative to GPT-4o for text tasks
Pricing
Open Weights
$0
- 123B parameter model
- Self-hosting
- Research use
API (Mistral Platform)
From $2.00/1M tokens
- Hosted inference
- Commercial use
- Function calling
API (Azure/AWS)
Marketplace pricing
- Cloud integration
- Enterprise support
- SLA guarantees
Benchmarks
| Benchmark | Mistral Large 2 | Competitor |
|---|---|---|
| MMLU | 86.7% | Llama 4: 87.1% |
| HumanEval | 88.4% | GPT-4o: 90.2% |
Technical Specs
Parameters
123B
Context Window
128K tokens
Modalities
text
Languages
Open Source
Yes
License
Mistral Research License (open-weight)
Related Models
Llama 4
Meta AI
Meta's next-generation open-weight LLM with MoE architecture and strong multilingual support.
Stable Diffusion 3.5
Stability AI
Open-source image generation model with local execution, full privacy, and community ecosystem.
CodeLlama 70B
Meta AI
Open-source code generation model specialized for programming tasks across multiple languages.
Gemma 3
Google DeepMind
Google's lightweight open-source model family for efficient on-device and edge deployment.