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

The new benchmark turns 114 marketing briefs into code-rendered videos; Claude Opus 5.5 leads by 93 points, GPT-6 Astra ties for second, and Gemini 4 Argon serves as the judge model.

9to5Google reports Google acknowledges Gemini 4 Argon ranks well behind Claude and GPT-6 on Android coding benchmarks.

In Antigravity, a single text prompt has Gemini 4 Argon model and rig assets in Blender, batch import them into Unity, and generate editor tools for wave physics, shaders and lighting.

A Base ecosystem daily recap notes Capminal integrated Gemini 4 Argon into OpenCAP with a 1M-token context window, among other trading and ecosystem updates.

patloeber highlights Luciano Martins' work using Gemini 4 Argon to speed up vLLM and Gemma, a developer-side performance effort.

A new Google DeepMind paper reports four Gemini 4 Argon coding agents self-organized with no human help, wrote 2,783 programs and killed the Ender Dragon in 27 hours 14 minutes.

A post lists per-million-token input/output pricing for GPT-6 Astra, Opus 5.5, Gemini 4 Argon and others, with Argon marked as a limited-time $2/$10.

A reply post that only says "According to Gemini Argon:" with an attached image, giving no context on the question or conclusion.

The author pairs Gemini 4 Argon with the transkun piano transcription model to analyze and visualize their playing, separating left and right hands from MIDI and video.

AI Studio Build works well with Gemini 4 Argon and adds one-click UI chips for making games, building Three.js 3D scenes and using DeepMind's science skills.

An image post shows a take from Gemini 4 Argon that the poster describes as crazy, with the details visible only in the image.

A short post saying only "beautiful stuff !! Gemini 4 Argon" with no details or evidence.