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

A tutorial combines Gemini 4 as main model with 3.8 Flash, a browser plugin and Google Flow, chaining each clip's last frame into the next, costing 120 Flow credits and 3% of weekly quota for a 30-second video.

A user describes Gemini Argon Lite's conversational style as savage, darkly humorous and argument-dismantling, expressing shock that this is only the Lite tier.

Using Argon in Antigravity, the author produced a five-act supercar teardown: wireframe parts assembling mid-air, a laser scan revealing paint, engine ignition and a midnight highway chase, taking 45 minutes for the Ferrari and 14 for the Lamborghini.

A user posts an image claiming Gemini 4 Argon is now available in Google AI Studio for the Pro plan, an unconfirmed access report.

A user says running Argon Extra High via API to review a 30-page paper cost more than $30, after finding Deep Review's twice-weekly limit too tight.

A video compares Gemini 4 Argon against Opus 5.5, GPT-6 Astra and Sonnet 5.5, asking viewers which model leads.

A short post claims Gemini Argon 4 performs unusually well at chess, without offering benchmark details.

A video demo shows Gemini 4 Argon rebuilding NBA game footage into a court map from a single prompt, with the author noting precision is still improving.

A user sums up mixed feedback on Gemini 4 Argon in Antigravity: praised for creative writing but Pro quota runs out in hours, with Google reportedly resetting limits.

Gemini 4 Argon is now available in AI Studio for Pro users, enabling faster prototyping of next-generation AI apps.

GPT-6 Astra won, with Gemini 4 Argon tied for 3rd–5th with Claude Opus 5.5 and Fable 5.1; the best model beat the strongest human bot only about 7%, and models that sparred early against tough bots scored higher.

A user shares a toy-factory video generated with Gemini 4 Argon and asks for thumbs up or down.