Google has pushed Gemini 4 Argon out of limited testing and into general availability. The model, identified in Google's API as gemini-4-argon, became broadly accessible from 13 October 2026, a fortnight after Google first showed it to a small group of cybersecurity partners on 30 September.
Google positions Argon as a flagship reasoning model, aimed at software engineering, mathematics, science and multi-step agent work. The headline spec is context: Argon supports a combined input-output limit of 1,048,576 tokens, letting developers feed in entire codebases or long document sets in one call. It also supports code execution, function calling, file search and URL context as built-in tools.
Pricing follows an introductory structure. Through 30 November 2026, Google charges $2 per million input tokens and $10 per million output tokens. After that date, rates rise to $4 and $20 respectively.
Access depends on which product you use. Developers reach Argon through the paid tier of the Gemini API and AI Studio. Inside the consumer apps, Gemini app and AI Mode access is reserved for Pro and Ultra subscribers, while Antigravity, Google's agentic coding environment, restricts Argon to Ultra accounts only.
The rollout is staged rather than instant: Google has said the model can take several hours to appear across every account, even once a plan is eligible. For teams already testing Argon under the earlier restricted access, general availability mainly changes who can reach it and from where, not what the model does.
Source: MarkTechPost
