Google's Gemini 4 Argon Takes Top Spot with Strong Scores

Google has unveiled Gemini 4 Argon, a new frontier model that beats GPT-6 Astra and Claude Opus 5.5 on 13 of 19 benchmarks in Google's comparison.

The model is initially limited to select vetted cybersecurity teams through Google's Fairwind Program for trusted cyber defenders.

Google plans to expand access first to paid API customers and Google AI Ultra subscribers, but has not announced a date.

Argon scored 53 at high reasoning, matching GPT-6 Astra at its maximum reasoning setting, according to Artificial Analysis's evaluation.

The model also reports 77.9% on DeepSWE v1.1, a benchmark for software engineering, alongside leading scores on tests of knowledge work, long documents, and reading charts and video.

Google announced API rates start at $2 per million input tokens and $10 per million output tokens, which will rise to $4 and $20 after the introductory period.