Board/Mistral/Devstral Medium 2507
Mi

Devstral Medium 2507

Mistral AI·Jul 10, 2025N/A

Open weights

Coding-agent specialist (tuned for scaffolds like OpenHands) that punches two weight classes up on real-issue resolution while staying cheap enough for always-on automation.

Status
Generally available
License
Open weights · Custom
Context window
128K
Max output
32K
Input · $/M
$0.4
Output · $/M
$2
Cached input · $/M
—
Blended · $/M
$0.80
Parameters
Undisclosed
Inputs
Text only

Benchmarks

This model has no published scores on the benchmarks we track yet. Specs and pricing below the fold are still fully sourced.

Reasoning & knowledge

  • GPQA Diamond

    no public score

  • MMLU-Pro

    no public score

  • Humanity's Last Exam

    no public score

  • ARC-AGI-2

    no public score

Coding

  • SWE-bench Verified

    no public score

  • SWE-bench Pro

    no public score

  • LiveCodeBench

    no public score

Agents & computer use

  • Terminal-Bench 2

    no public score

  • OSWorld 2.0

    no public score

  • BrowseComp

    no public score

Math

  • AIME 2026

    no public score

Known strengths

  • Agent-scaffold tuning
  • Permissive-adjacent licensing

Caveats

  • Single-purpose: pair with a general model for chat/writing

Primary sources

  • Mistral models
Compare this model →

More from Mistral

  • MiMistral Medium 3.1Aug 1, 2025 · 128K ctx