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

A task-by-task list of recommended AI models and products assigns complex reasoning, OCR, and video understanding to Gemini 4 Argon, arguing tool selection beats chasing one top model.

A roundup of ten examples built with Gemini 4 Argon in its first eight days, from full 3D worlds and playable games to interactive machines and physics simulations.

Theo posts that T3 Code users processed ~36.5 trillion tokens in the same window, 1.5x all global Gemini 4 Argon demand, tongue in cheek.

The author ran the same prompt through Gemini 4 Argon and Opus 5.5 to generate a white tiger crossing a palace floor, with no edits or cherry-picking, and asks readers to vote on which version belongs in a trailer.

rohanpaul_ai converts Gemini 4 Argon's 25 trillion tokens over 72 hours into roughly 96 million tokens per second, arguing the economy was intelligence-constrained.

A nine-provider student plan comparison says Google AI Pro student edition gives access to Gemini 4 Argon, priced at $8.99/month in the US with YouTube Premium bundled.

A user shares a screenshot mocking how Gemini 4 Argon picks numbers, an anecdotal hands-on observation.

A same-month year-over-year comparison of flagship models argues Gemini 4 Argon ranks second in long-horizon engineering with the industry's lowest 15% hallucination rate, while GPT-6 Astra is narrowly specialized and Opus 5.5 still leads on complex tasks.

A user ran the gemini-4-argon checkpoint in Arena on a voxel steampunk brass kraken prompt and called it their best output so far.

The author says Gemini 4 Argon on LMArena produced a clean interactive digital sketchbook, arguing the entry actually routes to a Gemini 4 Carbon checkpoint.

A user shares two Arena session links so others can compare outputs from gemini-4-argon-high and the gemini-4-argon Carbon checkpoint.

NovaXCode says Gemini 4 Argon once burned a weekly limit in an hour, and the stronger Gemini 4 Carbon might wipe it out in 10 minutes.