AI model by Mistral AI
Mistral Medium 3.5 for business
A newer open-weight Mistral model for agent and coding work, with image input, tool calling, structured outputs and a 256K window.
- Coding
- Open weights
- General purpose
- Vision and multimodal
- Open weights
- Reviewed on September 24, 2026
- Mistral AI
Capability tiersRelative, not benchmarks
Key facts about Mistral Medium 3.5
- Model ID at review
- mistral-medium-3.5
- Provider
- Mistral AI
- Main category
- Coding
- Open weights
- Yes, Modified MIT
- Last reviewed
- September 24, 2026
Fit
Where it fits and where it does not
Good at
-
Agent workflows with tool calling.
-
Coding and code review.
-
Image input with text.
-
Structured outputs.
-
Open weights for controlled hosting.
Not the right choice for
-
Very high volumes of simple tasks.
-
Teams that need a permissive Apache license, since this one is modified MIT.
-
Audio or video input.
-
Adoption without testing, since it is recent.
Use cases
Business use cases we would use it for
Coding assistants
Document work
Our notes
When we would choose it
Mistral Medium 3.5 was released in April 2026. Mistral describes it as a frontier-class multimodal model for agentic and coding use. It accepts images, supports tool calling and structured outputs, and has a 256K token context window. The weights are published under a modified MIT license.
It sits between the efficient Mistral Small 4 and the large, general Mistral Large 3, with a focus on agents and code. For teams building agents who want an open-weight option, it is a natural candidate to test.
Two practical notes. First, the license is a modified MIT license, not Apache 2.0, so read the modifications before building a product on it. Second, the documentation pages we checked showed slightly different model IDs, so confirm the exact ID before you write code.
The meters on this page reflect Mistral's positioning and our early view. We did not verify pricing at review time, so the price level is an estimate. Confirm current prices with Mistral before planning costs.
For coding agents, compare it with Codestral for completion work and with hosted leaders such as Claude Opus for complex changes. See coding models and AI copilots.
For an agent test, we would give each model the same set of real tasks with the same tools, such as looking up a record, updating a ticket and drafting a reply. Then we count finished tasks, wrong tool calls and time taken. Those three numbers say more about fit than any general ranking.
Before you commit
Things to check before you commit
-
License
The license is a modified MIT license. Read the modifications.
-
Pricing
We did not verify API pricing at review time.
-
Model ID
Confirm the exact API model ID in the current documentation.
-
Quality
Test against Large 3 and hosted leaders on your tasks.
Alternatives
Models to compare it with
Mistral Large 3
Mistral's open-weight general-purpose flagship under Apache 2.0, with image input, tool calling, structured outputs and a 256K token window.
Codestral 25.08
Mistral's low-latency coding model for code completion and generation, with fill-in-the-middle support, tool calling and a 128K window.
Qwen3-Coder
Qwen's open-weight coding models under Apache 2.0, from the efficient Qwen3-Coder-Next to the large 480B model, with tool use and long context.
Keep exploring
Solutions, services and guides
Related services
View all related services- AI agents AI that takes actions in your systems, such as qualifying leads or processing requests, with people checking the results.
- AI copilots AI helpers built into your product or internal tools that draft, summarize and suggest while people decide.
- AI Product Development AI agents, knowledge assistants, copilots and document automation built into the way your team already works.
- MVP development A first version with just enough features to put in front of real users and learn what to build next.
- Document automation Read invoices, forms, contracts and IDs, pull out the right fields and route them for review.
- AI workflows Step-by-step automations where AI handles the reading, sorting and drafting inside a process you control.
Solutions
View all solutions- Approval workflows Requests, approvals and reminders in one place, with clear owners and deadlines instead of long email threads.
- AI document processing Read forms, applications, IDs and statements, extract the fields you need and route each document for the right review.
- AI lead qualification Score, research and route every new inquiry so your team talks to the best leads first.
- Automated quoting Draft accurate quotes and proposals from a request, your price rules and past work, for a person to approve.
- Email triage Sort a shared inbox by topic and urgency, pull out the key details and draft replies for a person to send.
- AI sales research agent Short, sourced briefings on each prospect before a call, drafted by an AI agent from public information and your CRM.
Industries
View all industries- Real estate Listing sites, lead handling, CRM automation and document workflows for agencies, brokers and developers.
- Tax and compliance services Client portals, document checklists, case queues and deadline tracking for tax preparers and compliance firms.
- Legal Contract review assistants, knowledge search, intake and document automation for law firms and in-house legal teams.
- Professional services Client portals, proposal drafting, knowledge assistants and internal tools for consultancies, agencies and firms.
Case studies
View all case studiesGuides and articles
View all guides and articles- Where AI helps in development, and where it does not An honest map of the development tasks AI speeds up and the ones that still need experienced people.
- What is AI-accelerated development? How software teams use AI to plan, code, test and document faster, and where people still make every decision.
- What are AI agents and how do businesses use them? What AI agents are, real business uses, the controls they need and how to start with one safely.
- How to build an AI workflow with MCP servers Plan, connect and test an AI workflow that uses MCP servers, from a single task to a reliable process.
- AI chatbot vs AI agent What separates a chatbot that answers from an agent that acts, and when you need each.
MCP servers
View all mcp servers- GitHub GitHub's official MCP server lets AI read repositories, search code, manage issues and pull requests and, when allowed, change files.
- GitLab GitLab's built-in MCP server lets AI read projects, code, merge requests and pipelines and create commits and work items. In beta.
- Sentry Sentry's official MCP server lets AI search errors, events and traces, read issue details, get AI analysis and update issues.
- Context7 Upstash's Context7 MCP server gives coding assistants current, version-specific documentation for thousands of libraries.
- Everything The MCP project's Everything server is a test server that shows every protocol feature. For building and testing MCP clients only.
- Figma Figma's official MCP server gives AI design context, screenshots, variables and assets, and can now write to the canvas.
Glossary terms
View all glossary terms- AI agent An AI agent is software that uses a language model to plan steps, call tools and act toward a goal, instead of only answering one question.
- AI copilot An AI copilot is an assistant built into a tool people already use, which suggests, drafts and explains while the person stays in control.
- Claude Code Claude Code is Anthropic's agentic coding tool, which works in a developer's terminal, IDE, desktop app or browser to read code, run commands and make changes.
- Artificial intelligence Artificial intelligence is the broad field of building software that performs tasks that normally need human judgment, such as understanding language or images.
- Automation Automation is using software to perform a task or process with little or no human effort each time it runs.
- Function calling Function calling is a model feature that lets an AI model request a specific function, with structured inputs, for the application to run.
FAQ
Questions people ask us
Have a question that is not here? Ask us directly.
Yes, under a modified MIT license. Read the license terms before use.
Mistral positions it for agentic and coding work.
Yes. Mistral lists image input.