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

A Chinese post ties Gemini 4 Argon's perceived quality drop to mass student account removals and guesses Google is short on compute.

A roundup notes that a month after release, Opus 5.5 still ranks first despite new arrivals including GPT-6.1 Sol, Gemini 4 Argon and DeepSeek-V4.1.

A developer's project built with Gemini 4 Argon was featured on Google's official blog, shared with a screenshot.

A joke thread has DeepSeek V4.1 write code, Gemini 4 Argon flag two issues, GPT-6 Astra dispute Argon's review, Opus 5.5 side with Astra and jab Argon, ending with Argon apologizing.

An image post asks why a model comparison left out Gemini 4 Argon, reflecting community demand for Argon in head-to-head benchmarks.

An image post credits Google DeepMind and calls Argon a phenomenal model, offering positive hands-on feedback on Gemini 4 Argon.

A user reports the Gemini 4 Carbon checkpoint is still in Arena and started returning outputs again after a few hours of silence.

Privacy-focused AI platform Dot pitches its agent workspace and pay-per-call inference, claiming Gemini 4 Argon costs nearly 60% less than market pricing when credits are funded with $DOT.

A pricing comparison puts Kimi K3.1 Pro fourth globally and first in China, cheaper only than Fable 5.1, Opus 5.5 and GPT Astra, but pricier than Gemini 4 Argon, GPT 6.1 Sol and GLM 5.3.

The post lists Argon aiding plasma physics, five lab-verified enzymes, a missed newborn genetic abnormality and near-doubled quantum error-correction speed at over 100x lower cost.

A user who bought a cheap Pro membership finds Gemini 4 Argon only runs in Spark Beta with no model choice, hits version-number hallucinations and burns through usage limits fast.

A long post claims free Pro accounts get routed to a cheaper gemini-4-argon-trail branch while Ultra gets the original, and Antigravity itself runs at about 80% of AI Studio/Vertex quality.