Mistral: Mistral Medium 3.5 logo

MODELS

Mistral: Mistral Medium 3.5

by Mistral AI

modellead-sourceopenrouter-modelsmistral-aimultimodalopen-weightscodingagentslong-contextsource:mistral.aimistral-medium-3-5reasoningapi
Visit WebsiteDocumentation

Overview

Mistral Medium 3.5 is Mistral AI’s open v26.04 frontier-class multimodal model, listed with a 256k context window and API pricing for input and output tokens.

Details

Mistral Medium 3.5 is described in Mistral AI’s official model card as an open v26.04 frontier-class multimodal model with a 256k context window. Mistral’s pricing page lists it under API pricing with text-to-text, agentic, multimodal, coding, and reasoning capabilities, using the alias mistral-medium-latest. Mistral AI’s announcement describes Mistral Medium 3.5 as a public-preview flagship 128B dense model that merges instruction-following, reasoning, and coding, and says it powers Vibe and Le Chat Work mode. OpenRouter and the Hugging Face repository also describe it as a dense 128B instruction-following model supporting text and image inputs with text output.

When to Use

Use when you need a Mistral AI model for text-to-text, agentic, multimodal, coding, or reasoning workflows listed on the official pricing page. Use for long-context workloads where the official model card’s 256k context window is relevant. Use when text and image inputs with text output are needed, based on the OpenRouter and model-source descriptions.

Getting Started

  1. Review the official Mistral Medium 3.5 model card in Mistral Docs.
  2. Check the Mistral pricing page for the mistral-medium-latest alias and current input/output token prices.
  3. Use the Mistral Chat Completion API documentation to integrate with /v1/chat/completions.
  4. Review Mistral’s moderation and guardrailing documentation before production deployment.

Key Features

  • •Open v26.04 frontier-class multimodal model per the official model card
  • •256k context window listed in the official model card
  • •API pricing listed at $1.50 per million input tokens and $7.50 per million output tokens
  • •Model alias mistral-medium-latest listed on Mistral’s pricing page
  • •Described by Mistral AI as a public-preview flagship 128B dense model for instruction-following, reasoning, and coding

Capabilities

  • •text-generation
  • •image-input
  • •instruction-following
  • •reasoning
  • •coding
  • •agentic-workflows
  • •long-context

Last updated May 31, 2026