Skip to main content
AI Choice Engine

Mistral

Mistral Medium 3.5

Mistral Medium 3.5 is a Mistral model with 262K context, $1.50 in / $7.50 out per million tokens, last verified October 25, 2026. Open weights: yes. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

Mistral's merged 128B dense text/image model with a 262,144 checkpoint context, configurable reasoning, and use in Vibe remote agents and Le Chat Work mode.

Recorded profile verification October 25, 2026Individual provider fields may change

AI model specification details
SpecificationMistral Medium 3.5
ProviderMistral
TierFrontier
Context window262K
Max outputNot verifiedUnverified
Input / 1M tokens$1.50
Output / 1M tokens$7.50
WeightsOpen
Parameters128B dense
Reasoning levelsnone, high
Modalitiestext, image
LicenseModified MIT
API model idmistral-medium-3-5
ReleasedApril 28, 2026

Pricing tiers: Mistral first-party API $1.50/$7.50 per MTok; cached input $0.15. The 128B dense checkpoint is open weights under a Modified MIT licence with a monthly company-revenue restriction. The current model card is GA; April 28 is the docs/changelog release date, while the legal centre lists April 29 and the May 22 article describes public preview. These are the default Standard USD values in the retained pricing view. Batch/Priority and regional controls are offered, but their exact tariffs and eligibility are not established by that view; do not assume a universal discount or residence guarantee.

Publisher evidence

Published benchmark results

Each result keeps its source date visible. Dates refer to source publication or capture unless the source explicitly states when it measured the result. A missing benchmark is not treated as a zero.

Published benchmark results for Mistral Medium 3.5
BenchmarkScoreSource dateSource
Artificial Analysis Intelligence IndexMistral Medium 3.5142026-10-01Artificial Analysis · View source

Best for

  • EU-hosted frontier work
  • Open-weight self-hosting
  • Mistral Vibe agents

Watch out

Modified MIT denies licence rights if company/employer global consolidated revenue exceeds $20M in the preceding month; it covers derivatives and combined works. A separate commercial licence or Mistral-hosted service is then needed. None disables reasoning; high enables it. Use the corrected Transformers config: the publisher reports long-context degradation in older configs and derived GGUFs. A separate hard output cap is not verified. EU residency depends on the endpoint. Native guidance recommends temperature 0.7 and top_p 0.95 for high reasoning; generation examples are not hard output ceilings.

History

Mistral Medium 3.5 timeline

Release and status events for this model that passed two-source verification — dated, sourced, and linked.

  • Mistral Medium 3.5ReleaseApril 28, 2026

    Mistral docs/changelog record Medium 3.5 on April 28: a 128B dense open-weight model with Modified MIT terms, $1.50/$7.50 API rates and 256K context.

    Mistral docs changelog

Common questions

Mistral Medium 3.5

Answered from the verified figures on this page rather than general guidance.

How much does Mistral Medium 3.5 cost per million tokens?
Mistral Medium 3.5 lists base rates of $1.50 per million input tokens and $7.50 per million output tokens. Estimate both input and output token volume and any additional charges for your workload. Mistral first-party API $1.50/$7.50 per MTok; cached input $0.15. The 128B dense checkpoint is open weights under a Modified MIT licence with a monthly company-revenue restriction. The current model card is GA; April 28 is the docs/changelog release date, while the legal centre lists April 29 and the May 22 article describes public preview. These are the default Standard USD values in the retained pricing view. Batch/Priority and regional controls are offered, but their exact tariffs and eligibility are not established by that view; do not assume a universal discount or residence guarantee.
What is Mistral Medium 3.5's context window?
Mistral Medium 3.5 accepts about 262K tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is Mistral Medium 3.5 best for?
Mistral Medium 3.5 is a frontier tier from Mistral. It suits eu-hosted frontier work, open-weight self-hosting, mistral vibe agents. Modified MIT denies licence rights if company/employer global consolidated revenue exceeds $20M in the preceding month; it covers derivatives and combined works. A separate commercial licence or Mistral-hosted service is then needed. None disables reasoning; high enables it. Use the corrected Transformers config: the publisher reports long-context degradation in older configs and derived GGUFs. A separate hard output cap is not verified. EU residency depends on the endpoint. Native guidance recommends temperature 0.7 and top_p 0.95 for high reasoning; generation examples are not hard output ceilings.
Can I self-host Mistral Medium 3.5?
Mistral Medium 3.5 publishes open weights, but self-hosting depends on the licence, hardware footprint, quantisation quality, and serving stack. A hosted API is often cheaper until you have measured throughput and concurrency on your own hardware.