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

One user says they are eager for Gemini 4 Argon because they are tired of the strong AI tone in text written by other models.

A French-language post criticizes Google for being outmatched in its own ecosystem, saying Gemini 4 Argon would be unacceptable if it performs worse than Sol 6.1 or Opus 5.5.

A user posts that Argon is nearly here and expected to be Google's best model yet, while joking about whether it can translate French well.

StatsWire says Gemini 4 Argon is rolling out and some users have already received access.

The poster says Gemini 4 Argon's release is close and the timeline is flooded with screenshots, some real and some edited, making them nearly impossible to tell apart.

A skeptical post asks how unreleased Gemini Argon leads Arena with 1533 points, 21 ahead of Claude Opus 5.5, while Polymarket puts 80% odds it won't launch before November 1.

The poster says Gemini 4 Argon is finally rolling out to some users and looks forward to seeing tests on the timeline today.

A post says Gemini 4 Argon has begun a limited rollout to a small number of users and includes a screenshot.

A video shows Claude and Codex integrated into Google Workspace, and the poster jokes about getting justice for Gemini Argon.

A Japanese author reflects on the flood of Opus 5.5, GPT-6.1 Sol and Gemini 4 Argon posts, saying he was consuming AI news without producing results, and now picks models by task and daily condition.

A roundup of Argon leaks says the model has appeared in Google Cloud and as the default in the Antigravity CLI, with a 1M-token context, $2/$10 pricing and reported benchmark scores.

Gemini 4 Argon has quietly appeared in Google Cloud, with some people testing it in limited form as a silent rollout begins.