Mercury 3
Inception's diffusion language model provides long-context text generation, adjustable reasoning, tool calling and structured outputs through an API.
Mistral Large 4
docs.mistral.ai
A multimodal model for coding, document understanding and applications that call tools.
Visit website
Pricing
Paid
Available on
API, Mistral Studio
Category
Llm Models
What is Mistral Large 4?
Mistral Large 4 is a general-purpose multimodal model available in public API preview through Mistral Studio. Mistral positions it for coding, agent workflows and multimodal understanding, including enterprise applications.
Developers can use its one-million-token context window alongside document question answering, structured outputs and function calling. The model identifier is mistral-large-4.
Key features
Large context window
Accept up to one million tokens of context for applications working with substantial source material.
Document question answering
Use document Q&A through chat completion or conversation endpoints.
Structured responses and function calling
Request structured outputs and connect model responses to application functions.
Agent and batch integration
Access agents, conversations and built-in tools through their respective endpoints, with batching available separately.
Who is it for?
Developers building coding assistants
Evaluate the model for code explanation, implementation suggestions and repository questions.
Enterprise teams reviewing documents
Build question-answering workflows around lengthy business material.
Application teams connecting tools
Prototype assistants that request application functions and return structured results.
How to use Mistral Large 4
- Create a Mistral account and check access to the Large 4 preview in Studio.
- Create an API key and install the Python or TypeScript SDK.
- Select mistral-large-4 in a supported request.
- Submit a representative task and inspect the response before integrating it into your application.
Mistral Large 4 pricing
Paid
Usage-based API billing charges input, cached input and output tokens separately. Standard rates are distinct from batch, priority and regional pricing options.
Standard API
USD 1.36 per million input tokens; USD 0.14 per million cached input tokens; USD 4.18 per million output tokens
Metered token usage; no recurring subscription interval
Standard Mistral Large 4 API rates. Other processing tiers and regional options have separate pricing controls.
What to know before you use it
- The documented release is Public Preview, which should factor into deployment decisions.
- The launch announcement schedules downloadable weights for the end of October; it does not establish an available self-hosted release.
- Capabilities depend on the endpoint: built-in tools use agent or conversation endpoints, while batching uses the batch endpoint.
Frequently asked questions
Can I download the weights now?
Mistral's launch announcement promises weights by the end of October. It identifies the public preview API as the initial access route.
What do I need for the SDK quickstart?
A Mistral account, an API key and either Python 3.9 or later or Node.js 18 or later. Confirm Large 4 access for your account before integration.