SafetyPage

Lab system cards and frameworks · Model card · Google DeepMind

Gemini 2.5 Pro Model Card

Read at storage.googleapis.com ↗Read at sourceAll rights reserved

Outline · 21 pages

  1. Model Informationp. 2
  2. Model Datap. 3
  3. Implementation and Sustainabilityp. 3
  4. Evaluationp. 4
  5. Intended Usage and Limitationsp. 7
  6. Ethics and Safetyp. 7
  7. Frontier Safety Critical Capability Evaluationsp. 10
  8. CBRNp. 12
  9. Cybersecurityp. 14
  10. Machine Learning R&Dp. 17
  11. Deceptive Alignmentp. 19
  12. Frontier Safety: Summaryp. 20
  13. Appendix: Frontier Safety Correctness Testsp. 21