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

The author guesses Gemini 4 Flash won't cost much and may be usable within a Google Pro subscription, citing pricing of Luna, Haiku and Chinese Flash models.

PromptLLM posts an image of a controversial take produced by Gemini 4 Argon, circulating it as a community meme.

A user shares a set of 3D objects created with Gemini 4 Argon, calling the results delightful and worth browsing.

A poll asks which single model to keep among Grok 4.7, GPT-6.1 Sol, GLM-5.3, Claude Opus 5.5, Gemini 4 and Argon 4.

In a 780K+ human-vote benchmark of LLMs generating 3D objects from code, Gemini 4 Argon ranks first, followed by GPT-6 Astra.

A Lighthouse benchmark compares Claude Opus 5.5 and GPT-6 Astra on a lighthouse render with cost and time, and asks whether Gemini Argon should be tested next.

A Spanish post recounts how Wiz's Atlas agent with Gemini 4 Argon and GPT-5.6 Sol found decade-old PHP flaws enabling unauthenticated RCE, affecting WordPress, Drupal and Joomla.

Google DeepMind's Discovery team built a Gemini 4 Argon pipeline that simulates Turing machine traces, conjectures rules and proves them in Lean 4, finding 20 new potential cryptids and verifying 33 more machines.

A user recommends using Gemini 4 Argon to generate ad scripts, calling it worth trying for creative copywriting.

Replying to Tim Soret, banteg recommends Gemini 4 Argon, calling its ideas over the top with an interesting engineering brain, and says he used it to add an insane alternative rendering pipeline to a 2003 game.

A short video post where the author shares their personal impressions of Gemini 4 Argon.

Two critical PHP core vulnerabilities over a decade old were disclosed and patched, with Wiz and OpenAI saying ChatGPT 5.6 Sol and Gemini 4 Argon helped find them.