On October 13, 2026, Google made Gemini 4 Argon available outside its first, restricted rollout. Argon was announced on September 30 and went first to approved cybersecurity defenders through Google's Fairwind Program. Google's post on October 13 says "Google AI Pro and Ultra subscribers can now access Gemini 4 Argon" in the Gemini app. 9to5Google reports that the rollout is gradual and that Pro and Ultra subscribers also get it in AI Mode in Search.
Who gets it, and where
- Consumers: Pro and Ultra subscribers in the Gemini app and AI Mode. 9to5Google notes that last month, Gemini app access was limited to the Ultra tier.
- Developers: VentureBeat lists the paid Gemini API, Google AI Studio, the Antigravity agentic development environment, and eligible Gemini Enterprise subscriptions. 9to5Google reports that the Antigravity desktop app and CLI, AI Studio and Android Studio Canary require AI Ultra.
- Cyber partners: VentureBeat reports access has expanded to "hundreds of additional approved cybersecurity partners", some of whom get versions without certain cyber restrictions for authorised defensive work.
The price is introductory
Argon costs $2 per million input tokens and $10 per million output tokens until November 30, with cached input at $0.10 per million. VentureBeat: "Beginning December 1, the standard public API price doubles to $4 per million input tokens and $20 per million output tokens." That is the same as Claude Opus 5.5 and Meta's Muse Wonder, and the same as Claude Sonnet 5.5 and GPT-6.1 Sol during the introductory window.
What independent leaderboards say
The September launch came with Google's own benchmarks. Now there are outside numbers, and they disagree. VentureBeat reports that "As of October 12, Argon scores 68.90%" on the Vals Index, first, ahead of Claude Sonnet 5.5 at 67.04% and Claude Opus 5.5 at 66.97%. On the Artificial Analysis Intelligence Index, Argon scores 53, level with GPT-6 Astra and below Claude Opus 5.5 at 58. Sam Witteveen, CEO of Red Dragon AI and a VentureBeat host and partner, said in a video review shared with VentureBeat: "This is a really, really, really good model. But is it the best model out there? I actually don't think it is." He added: "I still don't think, for example, Antigravity is anywhere near as good as Claude Code or Codex."
The launch day's Hacker News thread passed 1,100 points and 800 comments. Several commenters praised Argon for non-coding questions, and one said Claude Code with Opus 5.5 had proved significantly better than Gemini Pro on their Python code analysis.
What remains uncertain. Google's post names only Pro and Ultra subscribers and gives no timeline for other tiers. The 1 million-token limit, which Google now describes as combined input and output, needs API configuration and a continuation protocol, and VentureBeat says Google has not clarified how it relates to the longest single uninterrupted response.
Analysis: The interesting number is the December 1 price. At $2/$10, Argon is priced like a mid-tier model while competing with flagships. At $4/$20 it is priced like Opus 5.5 and has to win on merit. Teams evaluating it now should budget at the December price. And because the two independent indexes disagree, run your own task-specific evals before you switch.
Google opened Gemini 4 Argon to AI Pro and Ultra subscribers, the paid Gemini API, AI Studio and Antigravity on October 13, at $2/$10 per million tokens until November 30 and $4/$20 from December 1; it leads the Vals Index but trails Claude Opus 5.5 by five points on Artificial Analysis's index.