Google DeepMind has released Gemini 4 Argon, its latest frontier AI model, making it available globally through the API in AI Studio, the Gemini platform, Google Antigravity for AI Ultra subscribers, and the Gemini app and AI Mode in Search for AI Ultra and Pro subscribers.

Among the use cases Google is promoting is the model’s ability to find and patch critical cybersecurity vulnerabilities, positioning Argon as a tool for security teams dealing with large, complex codebases. The company says the model excels at deep, multi-step work across coding, finance, legal, and scientific domains, and it claims the top spot on the Vals Index, a benchmark measuring model performance across finance, coding, legal, and tax tasks.

Pricing and availability

Argon launches at an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted 95%.

Engineering and research impact claims

Google reports internal teams have used Argon in its Antigravity environment to increase agentic code submissions by more than 35% while cutting code rollbacks by nearly a third. Specific examples cited include:

  • Chrome refactoring: Argon helped separate a roughly 6-million-line legacy folder of browser features from Chromium’s 40-million-line codebase, completing an architectural refactor across hundreds of files in weeks instead of months.
  • Agent Development Kit optimization: Working autonomously overnight, Argon cut in-memory session prep time by 40% in long sessions and sped up cold imports by 21% in Google’s open-source ADK framework.
  • xTB quantum chemistry package: Argon made the widely cited molecular modeling tool up to 2.3x faster at structure prediction and 1.8x faster at conformer search, with changes merged into the official repository.
  • Gemma inference speedup: Using Argon, Google achieved a 2.85x end-to-end inference speedup on an A100 GPU via vLLM, boosting throughput on Gemma 4 E2B from roughly 37,900 to over 108,000 tokens per second with no quality loss.

Google also says Argon is being applied to novel research problems in quantum computing, nuclear fusion, and life sciences, alongside improved spatial reasoning and visual coding capabilities used for 3D visualization and prototyping.

For security teams, the headline claim worth tracking is Argon’s use in vulnerability discovery and patching at scale. As with any AI-assisted code analysis tool, organizations evaluating Argon for security workflows should independently validate findings and patches before deployment, particularly in production environments.