Google DeepMind has made Gemini 4 Argon generally available. It is the first model of the Gemini 4 family, and Google says more models will follow. Argon is aimed at long, complex work in coding, finance, legal, cybersecurity and science, and Google reports it ranks first on the Vals Index.
Availability is broad: the Gemini API (AI Studio and the paid tier), Google Antigravity, the Gemini app and AI Mode in Search for AI Ultra and Pro subscribers, and Gemini for business. The API offers a 1M-token unified input-output limit, which Google calls industry-leading, together with a new continuation protocol for very long outputs.
Pricing is introductory: $2 per million input tokens and $10 per million output tokens, with cached input 95% off. From November 30, 2026 the price rises to $4 and $20.
- Google says it is stronger at spatial reasoning, mathematical precision, visual coding and writing, and works autonomously over long agentic tasks in Antigravity
- Reported figures include 87% on ERQA and a 0.7% attack success rate on the Gray Swan indirect prompt injection benchmark
- Google cites internal engineering results such as a large Chromium refactor and speedups in ADK Python and Gemma inference, which are Google's own claims
- 10,000 scientists can get the top Antigravity usage tier free for 6 months, and cyber defenders can apply to the Fairwind Program for a version without cyber guardrails
- Google describes a multi-layer safety stack, including an automated monitor that reads the agent's reasoning and actions and blocks workarounds in real time
All performance claims come from Google's announcement and have not been independently verified.
Summary by angular.love. The original source has the full details.
