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

A user reports Gemini 4 Argon's weekly limits just reset with 91% weekly and 99% of the 5-hour quota left, and plans to burn through it coding in Antigravity.

Two developers give positive reviews of Gemini Argon, with the poster speculating it targets specific architecture problems at Google scale.

Author says every ChatGPT prompt except Instant now counts against quota, and claims four Medium-thinking Gemini 4 Argon prompts can exhaust a Gemini Pro five-hour limit.

A user reports Gemini Argon build in AI Studio ran hundreds of calls without hitting a limit, while the reverse-proxied playground channel caps at 50 per day.

The author argues Fable 5.5 is delayed and Google had a perfect window with Gemini 4 Carbon, but keeping it from public access killed much of the hype.

A user who got hands-on with Gemini 4 Argon says they now use only Antigravity, admitting it may not be the best but is genuinely fun to work with.

Developer ThePrimeagen says Google cooked with Argon 4, the rumored Gemini 4 Pro codename.

A user describes Gemini 4 Argon as a cognitive exoskeleton and the new king of text consulting, blending agent and chatbot almost perfectly.

A user reports with a screenshot that Gemini Argon sees its own name as triggering content.

A user complains that Argon, reportedly Gemini 4 Pro, feels no different from Gemini 3 and doubts the new model brings real improvement.

The author argues Gemini 4 Argon beats other models at roleplay and writing, joking that trying it makes rivals feel bland.

A daily study-log post notes in passing that Gemini 4 Argon is now available in the Antigravity AI editor; the rest is exam prep.