AI Progress Tracker

    Mistral Medium · Mistral Medium 3 (mistral-medium-2505)

    Mistral · 7 May 2025

    Ask the Advisor about this model
    45from 5
    benchmarks

    Mistral Medium 3 is a proprietary multimodal model with text-and-image input, text output, and a 128,000-token context window, positioned for enterprise tasks including coding and document analysis. Mistral Medium 3.1 subsequently shipped; this profile remains specific to the May 2025 mistral-medium-2505 release.

    context: 128Ktext$0.40 / $2.00 per 1M tokens (in / out)

    Benchmark results

    Every figure links to where it was published. Compare like with like — harnesses and prompt setups differ between vendors.

    BenchmarkScoreSource
    GPQA DiamondPhD-level science questions59.5%7 May 2025View
    OTIS Mock AIME 2024-25Competition mathematics32.2%7 May 2025View
    FrontierMath0.3%7 May 2025View
    MATH Level 581.6%7 May 2025View
    LMArena TextHuman preference Elo from blind chat votes (not a percentage)1424 Elo15 Oct 2026View

    Weighted mean of 4 verified benchmark results: GPQA Diamond 59.5%, OTIS Mock AIME 2024-25 32.2%, FrontierMath 0.3%, MATH Level 5 81.6%.

    Capabilities

    • Understands text and image inputs within a shared conversation.
    • Supports code generation and software-development assistance.
    • Processes long documents within its 128,000-token context window.
    • Supports function calling for integration with external tools.
    • Supports structured output for application workflows.

    What improved

    • Adds image understanding compared with the original text-only Mistral Medium.
    • Expands the context window from the original Medium model's 32,000 tokens to 128,000 tokens.
    • Introduces lower published API prices than the original Mistral Medium: $0.40 per million input tokens and $2.00 per million output tokens.

    Known flaws & limitations

    • Model weights are not publicly released under an open-weight licence.
    • Native audio input and audio generation are outside its documented text-and-image interface.
    • Produces text rather than native image output.

    Risks, safety & security issues

    • Not yet confirmed by public reporting.

    Good fit for

    • Enterprise assistants working with text and image-based documents.
    • Coding assistance and software-development workflows.
    • Long-document summarisation and information extraction.
    • Tool-connected applications requiring structured text output.

    Poor fit for

    • Applications requiring publicly available open weights.
    • Native speech-to-speech or audio-generation applications.
    • Native image-generation applications.
    • Tasks requiring more than 128,000 tokens in a single context.
    Security & risk dashboardLatest vulnerabilities, advisories and real-world incidents for Mistral Medium

    References

    Profile last reviewed 10/18/2026 and re-checked automatically every few hours against the vendor's published documentation, papers and public leaderboards. Scores come from the cited benchmark results above — confirm limits and pricing against the provider's own documentation before building on them.