Sponsored by Byond Boundrys Consulting - Empowering Ideas, Delivering Results
Mistral AI Advanced

Mistral Medium 3.5

Mistral Medium 3.5 is a 256K-context multimodal model for coding and agents, with open weights and API pricing of $1.50/M input and $7.50/M output.

General Purpose Language ModelTextImage Paid
In plain English

What is this model and why does it matter?

Mistral Medium 3.5 is a 256K-context multimodal model optimized for agentic and coding workloads, with API access and open weights.

CodingAgentsMultimodal document workEnterprise assistantsOpen-weight deployment
Model overview

Mistral Medium 3.5: features, use cases and important details

Mistral Medium 3.5 is Mistral AI’s frontier-class multimodal model released April 28, 2026.

Verified model facts

Mistral documents a 256K context window, text/image input, function calling, structured outputs, agents and conversation APIs, plus open weights under a Modified MIT license.

Current status

It is generally available through the Mistral API at $1.50/M input and $7.50/M output.

Best fit

It is best for coding, enterprise agents, multimodal document work and organizations that also want an open-weight deployment path.

Limitations

Its context is smaller than newer 1M-token models, the license is modified rather than standard MIT and the model page does not state a knowledge cutoff or max-output limit.

Get started

How to use this model

  1. Create a Mistral API key or obtain the open weights.
  2. Call mistral-medium-3-5 for hosted inference.
  3. Provide text and image inputs when needed.
  4. Use function calling, structured outputs or agent/conversation APIs.
  5. Review the Modified MIT license before self-hosting.
Copy and try

Example prompts

  • Review this document and image and produce a structured decision.
  • Use functions to complete this agent workflow.
  • Analyze this codebase and recommend the safest implementation.
Capabilities

What it can do

  • 256K context
  • Multimodal input
  • Agentic workflows
  • Coding
  • Function calling
  • Structured outputs
  • Open weights
Best for

Practical use cases

  • Coding
  • Enterprise agents
  • Document Q&A
  • Multimodal analysis
  • Self-hosting
Pricing

What does it cost?

$1.50 per 1M input tokens and $7.50 per 1M output tokens on Mistral's API. The model is also released as open weights under a Modified MIT license.

Input$1.50 / 1M tokens
Output$7.50 / 1M tokens
Simple summaryCurrent Mistral API pricing is $1.50/M input and $7.50/M output; organizations can also deploy the open weights under the Modified MIT terms.

What stands out

  • 256K context
  • Open weights
  • Function calling
  • Structured outputs
  • Lower cost than many frontier models

Things to consider

  • Modified rather than standard MIT license
  • Less context than 1M-class models
  • Self-hosting still requires substantial hardware
Limitations

Important restrictions and trade-offs

  • Knowledge cutoff and maximum output are not stated on the model page
  • Open-weight license needs review
  • Can still hallucinate or make agent mistakes
SimplifyAITools verdict

Our editorial take

A practical frontier-class Mistral model for teams that want multimodality, agents and an open-weight deployment option without flagship pricing.

References

Primary sources

  1. Open source 1 ↗
  2. Open source 2 ↗
  3. Open source 3 ↗
Keep exploring

Related AI models