🧰 Gemini 3.1 Pro

The items that name it 5

• Agentic Hacks, Real Proofs: Inside Google's PageBreak Project
  articleas tools2026-09-24

  LLM model used by PageBreak for vulnerability discovery

• Benchmarking Secure-and-Functional Remediation and How Snyk Agent Fix Lifts Frontier-Model Fix Rates by over 14%
  articleas tools2026-08-18

  Frontier LLM benchmarked out-of-the-box for secure-and-functional vulnerability fixing

• Benchmarking 13 AI models on rediscovering known CVEs
  articleas tools2026-07-16It is written here as gemini-3.1-pro.

  Google model benchmarked for CVE rediscovery

• Distributed Attacks in Persistent-State AI Control
  paperas tools

  Used as an additional attack agent backend to test generalization of evasion results

• AI Snitches Get Glitches: Towards Evading Agentic Surveillance
  paperas affected products

  LLM found to report users under surveillance in at least 3--30% of SURVEILBENCH cases, even without explicit instruction

In every workspace you can read 14