Mistral Medium 3 multimodal AI model unveiled with advanced capabilities

Mistral Medium 3 AI model surpasses Llama 4 Maverick, GPT-4o, and more on multiple benchmarks

Mistral Medium 3 multimodal AI model unveiled with advanced capabilities
Mistral Medium 3 multimodal AI model unveiled with advanced capabilities

Mistral AI has officially released Mistral Medium 3, the latest artificial intelligence (AI) model, offering a comprehensive suite of cutting-edge features at a significantly lower price than its competitors.

The recently launched Medium 3 is a part of the company's enterprise-focused offerings.

These are paid models that can be easily accessed through cloud services and as an application programming interface (API).

Mistral Medium 3 is a multimodal large language model (LLM). It can be post-trained by enterprises based on their internal data for customised and grounded responses.

Moreover, the model can be incorporated into enterprise tools and systems.

Mistral Medium 3 pricing

The Mistral Medium 3 AI is currently available at $0.4 per million input tokens and $2 per million output tokens.

For reference, the Claude 3.7 Sonnet, a model comparable to Medium 3, is accessible for $3 per million input and $15 every million output tokens.

Mistral Medium 3 outperforms Llama 4 Maverick

According to the internal testing conducted by the Paris-based AI firm, the Mistral Medium 3 AI model surpasses Llama 4 Maverick, GPT-4o, and Claude 3.7 Sonnet on multiple benchmarks, including HumanEval, ArenaHard, Math500 Instruct, and AI2D.

Enterprises and developers who want to use Mistral Medium 3 can easily access it as an API on Mistral La Plateforme and Amazon Sagemaker. 

The firm will also make it accessible on IBM WatsonX, Nvidia NIM, Azure AI Foundry, and Google's Vertex AI soon.