Mistral’s New AI Model Offers Top Performance at a Great Price
▼ Summary
– Mistral released Mistral Medium 3, an efficient AI model priced at $0.40 per million input tokens and $20.80 per million output tokens, claiming it outperforms 90% of Anthropic’s Claude Sonnet 3.7.
– The model surpasses recent open models like Meta’s Llama 4 Maverick and Cohere’s Command A in AI benchmarks and is deployable on any cloud or self-hosted environments with four or more GPUs.
– Mistral, founded in 2023 and backed by over €1.1 billion, targets coding, STEM tasks, and multimodal understanding, with beta testing in finance, energy, and healthcare for customer service and data analysis.
– Mistral Medium 3 is available via Mistral’s API, Amazon’s Sagemaker, and will soon launch on Microsoft Azure AI Foundry and Google Vertex AI, with enterprise fine-tuning options.
– Mistral also launched Le Chat Enterprise, a corporate chatbot service integrating third-party tools like Gmail and SharePoint, with future support for Anthropic’s MCP standard.
French AI innovator Mistral has unveiled its latest breakthrough “Mistral Medium 3” a high-performance model delivering exceptional results at a competitive price point. Designed for efficiency without sacrificing capability, this new offering stands out in a crowded market by outperforming many established rivals while keeping costs manageable for businesses.
Priced at $0.40 per million input tokens and $20.80 per million output tokens, Mistral Medium 3 claims to match or exceed 90% of Anthropic’s Claude Sonnet 3.7 in benchmark tests while surpassing recent open models like Meta’s Llama 4 Maverick and Cohere’s Command A. For context, a million tokens translates to roughly 750,000 words—enough to dwarf Tolstoy’s War and Peace by over 160,000 words.
Flexibility is another key advantage. The model can be deployed across any cloud infrastructure, including self-hosted setups with as few as four GPUs. Mistral emphasizes that its pricing undercuts competitors like DeepSeek v3, whether accessed via API or self-managed systems.
Founded just last year, Mistral has quickly made waves in the AI space, securing €1.1 billion in funding from investors like General Catalyst. Its client roster includes heavyweights such as BNP Paribas, AXA, and Mirakl, leveraging Mistral’s models for applications ranging from customer support to complex data analysis.
Mistral Medium 3 shines in coding, STEM tasks, and multimodal comprehension, making it particularly appealing for sectors like finance, energy, and healthcare. Beyond its API—which allows enterprise users to fine-tune the model—the solution is now accessible via Amazon SageMaker, with plans to expand to Microsoft Azure AI Foundry and Google Vertex AI in the near future.
The launch follows March’s debut of Mistral Small 3.1, with hints of an even larger model arriving soon. Alongside this release, Mistral has introduced Le Chat Enterprise, a business-centric chatbot platform integrating its AI models with tools like Gmail, Google Drive, and SharePoint. Initially available in private preview, the service is now generally available and will soon support Anthropic’s MCP standard, joining industry leaders like OpenAI and Google in adopting this framework for seamless AI integration.
With its combination of performance, affordability, and adaptability, Mistral Medium 3 positions itself as a compelling choice for enterprises looking to harness cutting-edge AI without breaking the bank.
(Source: TechCrunch)
