Gemini 4 Argon: reviewers rate it below Claude and GPT-6 despite Google's release
Overview
Indigo says Google's Gemini 4 Argon returns the company to the top tier of models without surpassing rivals, and that its Antigravity app trails competitors by six months to a year, which holds back the model's performance.
He reports that Argon builds apps quickly at low reasoning levels but scores 21% on Vibe Check for complex legacy codebases, against 81% to 85% for Claude and GPT-6. He says it suits advanced users who choose their own tools as a partner for writing structure and ideas, or for quick prototypes and designs, but is not yet a primary tool.
Khazix says Argon became generally available on October 14, after an October 1 announcement limited it to the Fairwind cybersecurity program. He says it is hard for Argon to challenge Claude Opus 5.5 and GPT-6 Astra, citing an Artificial Analysis score of 53 that ties GPT-6 Astra. He reports that Argon costs $2 per million input and $10 per million output tokens during a promotion ending November 30, then $4 and $20. Using Antigravity requires the $100 Ultra plan, while the $20 Pro plan covers the Gemini app.
Written by AI from the articles below · updated Oct 13, 9:31 PM ET
Check the sources:
Developments
2 developments
- Oct 13, 9:02 PM ET · 1 articleGoogle opens Gemini 4 Argon to general availability after October 1 limited announcementKhazix: Gemini 4 Argon is fully open, but trails Claude Opus 5.5 and GPT-6 Astra in tests
- Oct 13, 7:54 PM ET · 1 articleindigo reviews Gemini 4 Argon and says it regained frontier tier but not ahead of rivalsindigo: Gemini 4 Argon matches top-tier models but lags in its app and coding review
Article timeline
The articles in this story. Times are ET.
Khazix@Khazix0918XGemini 4 Argon is fully open, but trails Claude Opus 5.5 and GPT-6 Astra in testsAIGoogle made Gemini 4 Argon generally available on October 14, after an October 1 announcement limited it to the Fairwind cybersecurity program. Khazix says it is hard for Argon to challenge Claude Opus 5.5 and GPT-6 Astra, citing an Artificial Analysis score of 53 that ties GPT-6 Astra. Argon costs $2 per million input and $10 per million output tokens during a promotion ending November 30, then $4 and $20, and it ranks first on Vals Index at 68.9.
indigo@indigoxXGemini 4 Argon matches top-tier models but lags in its app and coding reviewAIIndigo says Gemini 4 Argon brings Google back to the top tier of models but does not surpass rivals, and its Antigravity app trails competitors by six months to a year. In coding, Argon builds apps quickly and well at low reasoning levels, but scores 21% on Vibe Check for complex legacy codebases compared with 81% to 85% for Claude and GPT-6. Google has previewed automated safety classifiers, cloud execution and a max reasoning tier as upcoming updates.

Heat trend
- Current comparable heat
- 19
- Peak comparable heat
- 19Oct 13, 10:00 PM ET
- Change in past 24 hours
- –
The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.