Log in
← All models

Devstral 2

Balanced

Mistral's coding-and-agents specialist — built specifically for SWE-style agent tasks.

Mistral
Maker
83/100
Overall
70
Value score
$0.88
Blended /1M
262K
Context
Open
Weights
Pricing: $0.40 in · $2.00 out / 1M tokens · synced from OpenRouter · updated 2026-06-22

Scorecard

Output Quality7.8/10 · 71% community
Agentic Ability8.6/10 · 75% community
Speed8.2/10 · 73% community
Value for $8.8/10 · 76% community
Reliability8.7/10 · 76% community

Devstral 2 is Mistral's open-weight model tuned for software-engineering agents: codebase navigation, multi-file edits, and tool use, with a 262K context. A focused, affordable option for dedicated coding agents.

Strengths
  • Purpose-built for coding agents
  • Open weights
  • Good tool use
Trade-offs
  • Narrower than a general model
  • Behind frontier on hard reasoning

Agents that run Devstral 2 well

Dedicated open-weight coding agents.

Best model for ClineBest model for AiderBest model for Kilo CodeBest model for OpenCodeBest model for Continue

Devstral 2 FAQ

Is Devstral 2 good for AI agents?
Dedicated open-weight coding agents. On our editorial scoring it rates 8.6/10 for agentic ability and 7.8/10 for output quality. Purpose-built for coding agents.
How much does Devstral 2 cost?
Via OpenRouter, Devstral 2 is priced at $0.40 in · $2.00 out / 1M tokens — a blended rate of about $0.88 per 1M tokens for typical input-heavy agent use.
What is Devstral 2's context window?
Devstral 2 has a 262K-token context window. Mistral is the maker.
Is Devstral 2 open source?
Yes — Devstral 2 ships with open weights, so you can self-host it for privacy or run it through a hosted provider like OpenRouter. That openness is a big reason it's popular in open agent stacks.
What are the downsides of Devstral 2?
The main trade-offs: narrower than a general model; behind frontier on hard reasoning. It's strongest for dedicated open-weight coding agents.

Devstral 2 compared

Devstral 2 vs Sonnet 4.6Devstral 2 vs GPT-5.4Devstral 2 vs Gemini 3.5 FlashDevstral 2 vs GPT-5.3-CodexDevstral 2 vs Kimi K2.7 CodeDevstral 2 vs MiniMax M3Devstral 2 vs Qwen3.7 PlusDevstral 2 vs GLM 4.7

The model is half the story — the agent is the other half

The model picks the moves; the agent runs the loop, the tools, and the guardrails. Once you've chosen a model, see which agent gets the most out of it.

Compare AI agents →