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

A developer says testing Gemini 4 Argon via Devin's CLI and desktop app is ridiculously cheap, and now uses SWE-2 as a daily driver model.

A commenter argues neither Gemini 4 Argon nor Opus 5.5 generates video, so the effect is likely code rendered live in the browser.

The post claims Gemini 4 Argon agent swarms uncovered a 12-year-old 9.2-CVSS zero-day in core PHP (PHP2Shell / CVE-2026-109618), and the author built an interactive 3D attack-vs-exploit simulation in a single HTML file.

A developer demos MicroDuck, a home robot planned with Gemini 4 Argon and trained on DGX Spark, detecting falls without sending camera frames to the cloud.

A user gave the same prompt to Gemini 4 Argon and Claude Opus 5.5 to generate a prowling white tiger 3D scene, with strong fur, mist and lighting despite neither being a video model.

Google's CEO said Gemini 4 Argon processed nearly 25 trillion tokens in its first 72 hours, triple the prior generation, forcing staged access for Pro/Ultra and API users.

Given the same prompt of a white tiger stalking a palace, Gemini 4 Argon delivers a cold, camera-staring killer while Claude Opus 5.5 roars like a furious beast king.

A poll asks readers to pick the best coding model among GPT-6.1 Sol, Claude Opus 5.5, Gemini 4 Argon and Grok 4.7.

A user argues Gemini 4 Argon barely had its launch moment before Claude Duet arrived, and that viral hype means nothing if the model isn't useful afterward.

A user shares 3D output from an updated Carbon checkpoint, saying it is noticeably better than Gemini 4 Argon at 3D though still not perfect.

An image post showcases Gemini 4 Argon with a caption saying it has zero chill, though the specific test isn't described.

A user reports a new Gemini 4 Carbon checkpoint named gemini-4-argon in Arena, producing the most accurate heart shape seen so far from a Three.js prompt.