AI Progress Tracker

    GPT-4.1 Nano · gpt-4.1-nano-2025-04-14

    OpenAI · 14 Apr 2025

    Ask the Advisor about this model
    28from 6
    benchmarks

    GPT-4.1 nano is a compact OpenAI API model designed for low-cost, low-latency tasks, with text and image input and text output. It supports long-context processing, structured outputs and function calling, but its modest coding benchmark performance makes it a poor fit for demanding autonomous software engineering.

    context: 1Mimagetextfile$0.10 / $0.40 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 questions48.9%14 Apr 2025View
    OTIS Mock AIME 2024-25Competition mathematics28.9%14 Apr 2025View
    FrontierMath1%14 Apr 2025View
    SimpleQA VerifiedShort factual questions (hallucination resistance)6%31 Aug 2026View
    MATH Level 570%14 Apr 2025View
    Aider PolyglotCode editing across six programming languages8.9%14 Apr 2025View

    Weighted mean of 6 verified benchmark results: GPQA Diamond 48.9%, OTIS Mock AIME 2024-25 28.9%, FrontierMath 1%, SimpleQA Verified 6%, MATH Level 5 70%, Aider Polyglot 8.9%.

    Capabilities

    • Accepts text and image inputs and generates text responses.
    • Supports a context window exceeding one million tokens.
    • Supports function calling for application-controlled tool use.
    • Supports structured outputs for schema-constrained responses.
    • Designed for fast, inexpensive tasks such as classification and autocomplete.

    What improved

    • Expands the context window substantially compared with GPT-4o mini.
    • Lowers standard API input and output token prices compared with GPT-4o mini; listed prices are USD per million tokens.
    • Outperforms GPT-4o mini on GPQA and Aider Polyglot in OpenAI’s launch evaluations.

    Known flaws & limitations

    • Modest Aider Polyglot performance limits suitability for demanding code-editing work.
    • The dated model can issue multiple calls to the same tool when parallel tool calling is enabled; OpenAI recommends disabling parallel tool calls.
    • Its June 1, 2024 knowledge cutoff means later information requires external grounding.
    • Does not support native audio or video input, or image and audio generation.

    Risks, safety & security issues

    • Repeated tool calls can duplicate external actions unless applications enforce idempotency and validate requests.
    • Incorrect answers or generated code can cause downstream errors when accepted without independent validation.
    • Time-sensitive decisions can be misled by stale information when applications do not provide current sources.

    Good fit for

    • High-volume text classification and routing.
    • Low-latency autocomplete and short-form text generation.
    • Schema-constrained extraction from supplied documents.
    • Lightweight tool-assisted workflows with validated, idempotent actions.

    Poor fit for

    • Demanding autonomous repository-level coding.
    • High-stakes decisions without independent verification.
    • Current-events questions without up-to-date external sources.
    • Applications requiring native audio or video processing.
    Security & risk dashboardLatest vulnerabilities, advisories and real-world incidents for GPT-4.1 Nano

    References

    Profile last reviewed 10/27/2026 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.