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

Aria's model API now lists Gemini 4 Argon at a lowest route price of $0.22 per million input tokens and $1.10 per million output tokens, about 89% below Google's standard price, while gemini-3.6-flash and gemini-3.7-flash routes now point to Gemini 3.8 Flash.

Argon reportedly watched a single human cornhole video, wrote its own physics simulator, trained a throwing policy and transferred it to the Apollo humanoid for 7 hits in a row, cutting weeks of robotics work to hours.

The author says older Gemini models sounded human but that Gemini 4 Argon has caught "GPT syndrome," becoming indecisive, hesitant and afraid to take a stance.

A user lists the flagship models the US and China are going all in on, including Gemini 4 Argon, GPT-6 Astra and Claude Opus 5.5 on the US side and DeepSeek-V4.1 and Kimi K3 on the Chinese side, and asks where the rest of the world is.

Google's pricing page shows Argon launching at half the price of Opus 5.5, returning to the same $4/$20 rate on December 1; the poster says the 48 days is exactly enough time to rebuild your stack on Gemini.

Google released Gemini 4 Argon globally in the paid Gemini API and platform, the Gemini app and Search for Pro and Ultra subscribers, and Antigravity for Ultra subscribers, with the model placing 5th on the Arena Overall leaderboard.

The user finds Argon smarter, steadier and more reliable but sometimes stiffly literal about instructions, still going off-script in Gem roleplay and offering to move to concrete actions after reading files.

Riding the Gemini 4 Argon launch hype, a studio removed the glasses from its own Gemini mascot as a fun tie-in update.

A Turkish user says they are so disappointed in Google that they don't want to try Gemini 4 Argon, and asks those who have used it to help change their view.

A post claims Google settled on the name Argon very late, and that during secret benchmarking it used the name gemini-3.8-flash while media were still guessing it was Gemini 4 Pro.

On a game development benchmark, Muse Wonder cost $3.45 against Gemini 4 Argon's $29.71; the poster calls the 9x cost a significant tradeoff but says a playable game is worth far more.

A video comparing the same prompt says Opus produced a stretchy jelly effect for $7.20 while Gemini 4 Argon produced a plastic look for $22, raising doubts about its hype.