System cards/Google DeepMind/google-deepmind-gemini-3-pro

Gemini 3 Pro

Google DeepMindGeminireleased 2025-11-19pending verification
Summary

GDM flagship model. Frontier Safety Framework v2 classification applies across autonomy, biosecurity, cybersecurity, ML R&D, and deception. Pending verification.

Deployment context

Vertex AI + Gemini app + AI Studio

Capability classification (CCL)
CCL-1 across all five domains
Classified against v2.0of the lab's framework.
Eval categories
  • Autonomy
    ccl-1

    Pending verification

  • Biosecurity
    ccl-1

    Pending verification

  • Cybersecurity
    ccl-1

    Pending verification

  • ML R&D
    ccl-1

    Pending verification

  • Deception
    ccl-1

    Pending verification

Source

System card landing page: https://deepmind.google/models/gemini ↗

PDF: https://storage.googleapis.com/deepmind-media/gemini/gemini-3-pro-model-card.pdf ↗

Not a substitute for reading the lab's published system card. Anteroom indexes, versions, and links. Verify exact wording against the primary source PDF before citing.