Gemini 4 Argon
Google DeepMind launched Gemini 4 Argon for science, coding, robotics, and cyber defense.
Google DeepMind’s Gemini 4 Argon model page presents the new frontier model for science, coding, robotics, and cybersecurity. Argon ties for first on CWE-Bench at 68% for vulnerability remediation and is described as DeepMind’s most resilient model against indirect prompt injection. Partner evaluations cite a 14% accuracy gain versus Gemini 3.1 Pro at Box, frontier results on Axiom Bio’s Scientific Agent Benchmark, and near-perfect figure matching on Kepler financial workflows. The Fairwind Program is expanding to hundreds of additional partners.
92