These are apparently the leaked benchmarks for Gemini 4 Pro, the next version of Gemini.

If true, Gemini outperforms GPT-6, Astra, and Claude Fable 5.1 based on these benchmarks. It could be one of the greatest comebacks.

And with the news that Gemini has joined the hacking club, It may be true.

I’m looking forward to trying it out in Google Workspace Once it rolls out.

October Update:

Google has announced Gemini 4 Argon, with these benchmarks.

Auto-generated description: Performance comparison of Gemini 4 Argon against GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 across various benchmarks revealing varying degrees of success in knowledge work, agentic coding, ML engineering, science and math, long context handling, computer use, multimodal understanding, and cybersecurity.