Llama 4 Maverick 17B 128E Instruct FP8 profile

    Llama 4 Maverick 17B 128E Instruct FP8 — security & risk dashboard

    Known weaknesses, safety and security advisories, and documented real-world incidents for Llama 4 Maverick 17B 128E Instruct FP8 — collected automatically and re-checked every few hours, so teams building on it can see what to design around.

    Last checked 19 Oct 2026

    Ask the Advisor about this model
    Known flaws

    5

    Security risks

    4

    Real-world incidents

    0

    Advisories logged

    0

    Security, safety & abuse risks

    • Harmful or biased outputs remain possible despite instruction tuning and safety mitigations.
    • Adversarial instructions or untrusted retrieved content can challenge application safeguards; tool permissions require independent enforcement.
    • Generated text and code can be misused for deception or other prohibited activities.
    • Sensitive information in input documents requires deployment-level access controls, retention policies and output handling.

    Known flaws & limitations

    • Can produce inaccurate or fabricated information; important outputs require independent verification.
    • Generates text, not native image, audio or video outputs.
    • The complete model has approximately 400 billion parameters; 17 billion active parameters do not imply a 17-billion-parameter storage footprint.
    • The published benchmark table reports Maverick results, not a separate comparison establishing FP8 parity across serving configurations.
    • Deployment is governed by the Llama 4 Community License and Acceptable Use Policy, including restrictions affecting multimodal licensing in the European Union.

    Security advisories & risk updates

    No risk advisories logged since this profile started being tracked. New ones appear here automatically.

    What this means if you are building on it

    • High-stakes medical, legal or financial decisions without qualified review.
    • Applications requiring native image, audio or video generation.
    • Memory-constrained deployments sized for a dense 17-billion-parameter model.
    • Autonomous workflows that treat generated instructions as authorization for sensitive actions.

    Everything here comes from published documentation, advisories and reporting, with links to the original sources. It is not a complete vulnerability database — always check the provider's own security notices before shipping.