Gemini 4 Argon wins 14 of 19 tests against Claude and GPT-6
A comparison scorecard shows Gemini 4 Argon leading on legal, finance and long-document tasks, while GPT-6 Astra wins software engineering and computer use and Claude Opus 5.5 wins terminal coding.
🚨 BREAKING: Google just took the crown back. Gemini 4 Argon beats Claude and GPT-6 on 14 of 19 tests, and anyone with Gemini Pro can use it today. Where Argon wins: → Legal agent work, 19.6% vs Claude Fable 5.1’s 6.7% → Business automation, 51.3% vs Opus 5.5’s 42.5% → Finance agent work, 65.4% vs Fable 5.1’s 58.9% → Long documents up to 1M tokens, 84.2% vs GPT-6’s 71.8% Where Claude Opus 5.5 wins: → Terminal coding, 66.4% vs Argon’s 57.4% → ML engineering, 49.3% vs 45.3% Where GPT-6 Astra wins: → Software engineering, 65.5% vs Argon’s 55.0% → Computer use, 72.6% vs 69.2% Google wrote the scorecard. Benchmaxxing or is this the real deal?