← Models

Devstral 2

Mistral AI · Devstral

Deprecated Devstral 2 coding-model entry, retained for pinned integrations now routed toward Medium 3.5.

Type
language
Context
256K tokens
Max Output
256K tokens
Status
deprecated
Input
$0.4/1M tok
Output
$2/1M tok
API Access
Yes
License
Modified MIT
coding agentic open-weights fill-in-the-middle function-calling long-context
Released December 2025 · Updated June 20, 2026

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on June 20, 2026.

Devstral 2 is Mistral AI’s December 2025 coding model for agentic software work. It sat in the part of the lineup aimed at developers who wanted strong code-generation and code-editing behavior, model portability, self-hosting options, and lower-level control over the stack.

As of June 20, 2026, Mistral’s current docs mark Devstral 2 as deprecated with Mistral Medium 3.5 as the replacement. Keep this page for pinned devstral-2512 deployments, historical comparisons, and migration planning rather than as a fresh recommendation.

Capabilities

Devstral 2 was positioned for real software tasks: repository exploration, bug fixing, multi-file implementation, structured tool use, and edit-heavy coding flows. Official Mistral docs still list support across chat completions, function calling, structured outputs, predicted outputs, agents and conversations, built-in tools, prefix, batching, and fill-in-the-middle generation.

That combination is still relevant for coding. Fill-in-the-middle matters for editor-style completion and patching, while structured outputs and function calling matter for agent frameworks and coding CLIs that need reliable tool invocation rather than freeform prose.

Technical Details

Mistral’s official Devstral 2 page publishes a 256K context window and identifies the model as devstral-2512. The same page lists May 22, 2026 as the deprecation date and Mistral Medium 3.5 as the replacement. The docs do not foreground a separate completion ceiling in the same way some providers do, so this profile records maxOutput as the same 256K snapshot limit for repo comparability rather than as a distinct provider-published cap.

The official feature matrix also shows broad product-surface compatibility instead of a narrow “completion only” release, which is a strong signal that Mistral expects Devstral 2 to be used in agents, tooling layers, and coding assistants rather than only direct API prompting.

Pricing & Access

This repo keeps the last verified pricing values for historical comparability:

  • Input: $0.40 per 1M tokens
  • Output: $2.00 per 1M tokens

Check Mistral’s current pricing and deprecation tables before sending new traffic to the old endpoint, because the hosted route and self-managed weights may have different practical timelines.

Best Use Cases

Use Devstral 2 only when you are maintaining a pinned coding-agent integration or comparing older Devstral behavior. For new coding agents, start with Mistral Medium 3.5 if you want Mistral’s current featured agentic coding route.

Teams still need review discipline or a separate high-judgment reviewer model for risky migrations and security-sensitive work; a newer replacement does not remove that requirement.

Comparisons

  • Mistral Medium 3.5 (Mistral AI): Current replacement in Mistral’s docs and the better starting point for new Mistral coding-agent evaluations.
  • Mistral Small 4 (Mistral AI): Cheaper current Mistral route for lighter coding and assistant workloads.
  • GPT-5.4 (OpenAI): Proprietary frontier generalist with stronger product surfaces; Devstral 2 remains relevant mostly for old pinned Mistral deployments.