Mistral AI vLarge 2
Mistral AI vLarge 2: Mistral AI is a frontier AI platform for building tailored AI systems with open source models and Le Chat assistant.
Last verified:
What is Mistral AI vLarge 2?
Mistral AI vLarge 2: Mistral Large 2 is Mistral AI's flagship 123 billion parameter large language model designed for enterprise-grade AI applications. It delivers state-of-the-art performance in reasoning, code generation, mathematics, and multilingual tasks with a 128,000-token context window. The model is optimized for single-node inference and excels at complex tasks requiring significant reasoning capacity.
Key features include native function calling capabilities, structured JSON output generation, support for 80+ programming languages (Python, Java, C, C++, JavaScript, Bash), advanced mathematical reasoning with 93% GSM8K accuracy, and 84% MMLU benchmark performance. It supports text-to-text tasks, multimodal inputs, structured output formatting, and streaming responses. The model has been fine-tuned to reduce incorrect outputs and better recognize when it lacks sufficient information.
Mistral Large 2 is designed for businesses, researchers, and developers looking for high-performing AI solutions. It targets enterprises needing sophisticated AI for code generation and debugging, data analysis and research, document synthesis, multi-step task scheduling, API integration, and multilingual applications. The model is suitable for single-node deployment scenarios and long-context applications requiring extensive reasoning.
Mistral AI vLarge 2 pricing
Pricing model: Freemium
Mistral Large 2 costs $3.00 per million tokens for input and $9.00 per million tokens for output. Batch processing gets a 50% discount. The Free tier provides limited access to Mistral models including chat and search. Pro plan is $14.99/month offering 6x more messages and web searches, all-day coding in CLI/IDE/web, and 40x more image generations. Team plan is $24.99/user/month with up to 30GB storage per user, domain verification, and data export. Enterprise plan includes custom models, custom agents, custom workflows, audit logs, SAML SSO, and white-label options. Education plan for students is $5.99/month. API pricing includes Mistral Large at $2/M tokens input and $6/M tokens output, Mistral Medium 3.5 at $1.5/M input and $7.5/M output, and Mistral Small 4 at $0.1/M input and $0.3/M output.
Mistral AI vLarge 2 pros
- 123 billion parameters for enterprise-grade performance
- 128K token context window for long-context tasks
- 84% MMLU benchmark accuracy, setting open model standards
- 92% HumanEval score for code generation
- 93% GSM8K accuracy for mathematical reasoning
- Supports 80+ programming languages natively
- Native function calling capabilities without extra setup
- Structured JSON output generation built-in
- Optimized for efficient single-node inference
- Multilingual support across numerous languages
- Fine-tuned to reduce incorrect or irrelevant outputs
- Better recognizes when information is insufficient
- Concise and focused output generation
- On par with GPT-4o and Llama 3 405B for coding
- Extensive training on large code volumes
- Streaming responses supported
- Text-to-text and multimodal capabilities
Mistral AI vLarge 2 cons
- $3 per million tokens input costs higher than small models
- $9 per million tokens output pricing expensive for volume
- 123B parameters require significant computational resources
- Single-node inference needs very expensive hardware
- No open-source Apache 2.0 license for commercial use
- Research License restricts some commercial derivatives
- Higher cost per token compared to Mistral Small
- Not optimized for edge or lightweight deployments
Frequently asked questions about Mistral AI vLarge 2
What is Mistral Large 2?
Mistral Large 2 is Mistral AI's flagship 123 billion parameter large language model with a 128,000-token context window. It delivers state-of-the-art performance in reasoning, code generation, mathematics, and multilingual tasks. The model is designed for single-node inference and achieves 84% on MMLU, 92% on HumanEval, and 93% on GSM8K benchmarks.
How much does Mistral Large 2 cost?
Mistral Large 2 costs $3.00 per million tokens for input and $9.00 per million tokens for output. Batch processing receives a 50% discount. This pricing applies to the API version of the model.
What programming languages does Mistral Large 2 support?
Mistral Large 2 supports 80+ programming languages including Python, Java, C, C++, JavaScript, and Bash. It excels at code generation, debugging, refactoring, and writing code across multiple languages with performance on par with GPT-4o and Llama 3 405B.
What is the context window size of Mistral Large 2?
Mistral Large 2 has a 128,000-token context window (128K tokens), the same size as Llama 3.1. This allows it to handle complex tasks requiring significant reasoning capacity and long-context applications.
Does Mistral Large 2 support function calling?
Yes, Mistral Large 2 has native function calling capabilities built-in. It also supports structured JSON output generation, making it versatile for integrating with external tools and APIs without additional setup.
What benchmarks does Mistral Large 2 achieve?
Mistral Large 2 achieves 84.0% on MMLU (5-shot), 92% on HumanEval for code generation, and 93% on GSM8K for mathematical reasoning. These benchmarks set new standards for open models in performance and cost efficiency.
Can I deploy Mistral Large 2 on my own infrastructure?
Yes, you can self-host Mistral models anywhere. Open-weight models like Mistral 7B are Apache 2.0 licensed for research and individual use, but commercial deployments require a Mistral license with separate terms for derivatives and production use. Enterprise customers can deploy on-premises, in private cloud, or on Mistral Cloud with full data residency.
What is the difference between Free and Pro plans?
Free provides limited access to Vibe with latest models, chat, search, creation, and Vibe for code for agentic coding. Pro at $14.99/month offers 6x more messages and web searches, complex task capabilities, all-day coding in CLI/IDE/web, 40x more image generations, and chat/email support.
Does Mistral Large 2 support multilingual tasks?
Yes, Mistral Large 2 has enhanced multilingual support spanning numerous languages. It demonstrates strong performance across multilingual tasks and is designed for domain-specific, transparent, and multilingual reasoning applications.
What enterprise features does Mistral offer?
Enterprise plans include custom SLAs, dedicated support, private deployments, custom models, custom agents, custom workflows, audit logs, SAML SSO, white-label options, and custom data residency. Enterprise APIs offer regional data processing controls, system-level SLAs, increased rate limits, and premium support.