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

Users report that after a few days on a free Google AI Pro trial, Gemini 4 Argon and Claude Opus 5.5 become available in Antigravity.

A user ran the same Arena task across Gemini 4 Carbon (new Argon), Kimi K3.1, Qwen 4 and Claude Opus 5.5, and favored Gemini.

A user reports Gemini 4 Argon is available on Pixel Pro accounts, possibly differing from the paid tier, and is running tests to share results.

A blogger benchmarks MiniMax M3.1 Flash, GPT-6.1 Sol, Gemini 4 Carbon and Kimi K3.1 on building an animated 3D Lightning McQueen, with time and token costs.

A four-stage Recon/Hunter/Critic/Ledger prompt for exhaustive repo security audits with isolated subagents, recommended for Gemini 4 Argon or GPT-5.6 Sol.

A user currently on Opus 5.5 asks the community for a single reason to switch to Gemini 4 Argon.

The roundup says the PHP2Shell flaw was found with GPT-5.6 Sol's help and independently reproduced by Gemini 4 Argon, alongside Meta earnings and the Muse Spark Open release.

A user says Antigravity now uses Gemini 4 Argon with a weekly quota and is testing how long it lasts.

A post says Gemini 4 Argon is rolling out for complex tasks and asks what first real project users would hand it to build from scratch.

Some Jio and student users report Antigravity only offers Gemini 3.8 Flash, with Gemini 4 Argon and Claude Opus 5.5 unavailable despite the same 'Google AI Pro' label.

Users joke about inconsistent Google subscription tiers: some free Pro accounts get Opus 5.5, Sonnet 5.5 and Gemini 4 Argon, while others find their restored membership still limited to Opus 4.6.

A user lists rumored upcoming models — Kimi K3.1, Qwen 4, Fable 5.5, Gemini 4 Carbon and GPT 6.1 Astra — and asks which people are most excited for.