Gemini 4 Pro / Flash: benchmarks, hands-on tests and leaks

A user says Gemini 4 Argon is now live for Pro users in the Gemini app on mobile and desktop, and had it build a simple landing page as an early test.

A prompted robot video-to-simulation example demonstrates real2sim2real for robotics, linking to Google's developer blog page on Gemini 4 Argon for robotics.

The official GeminiApp account confirmed in a reply that Gemini 4 Argon has been released.

Router announced that Gemini 4 Argon is live on its platform, letting users call it with their own Google API key, with a demo video attached.

A user says tomorrow is a big day, thanks people for sharing their Argon tests, and jokes that they won't get a Google subscription for it.

A user says Gemini 4 Argon is now available to Pro subscribers, but only through the Gemini app, and is using it to build a project.

The author tested Gemini 4 Argon in High mode in Cursor, saying it feels completely different from Gemini 3.8 Flash, cost about $4 for 15 minutes, and ranks in his personal top four for 3D modeling quality.

Argon moves from its September 30 cybersecurity rollout to developers and enterprises, with Pro and Ultra access and introductory API pricing of $2/$10 per million tokens rising to $4/$20 in December.

One item in a daily crypto newsletter says Google launched its most powerful AI model, Gemini 4 Argon, alongside BTC, ETH and other market updates.

A post says Gemini 4 Argon can currently be used free via v0.app with Google sign-in monthly credits, or by selecting it on the Experiential Labs platform.

Hassan notes that Google AI Pro users can chat with Argon all day in the Gemini app, but coding with it in Antigravity requires Ultra, while API developers get it everywhere.

A joking post in which the user says they just woke up from a dream that Google released Gemini 4 Argon, calling it a terrible dream.