"No. 1 in the world, and Pro, Ultra and API users can't use it": Google's Gemini 4 Carbon scores 59 on the AI index, ahead of Opus 5.5

"Pro users: can't use. Ultra users: can't use. API developers: can't use. Artificial Analysis: world number one. Google: thanks for watching." An X user posted that on Oct. 28 about Google's Gemini 4 Carbon, released the same day. The post collected 784 likes.

Google's blog called Carbon its "strongest model," arriving two weeks after Gemini 4 Argon on Oct. 13. CEO Sundar Pichai announced it on X, where the post passed 22,000 likes. The headline result is first place on the Artificial Analysis Intelligence Index v5.0 with 59 points. VentureBeat reports Claude Opus 5.5 at 58, OpenAI's GPT-6 Astra at 57 and Argon at 50.

Introductory API pricing is $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 later, Google said. Anthropic's pricing page lists Opus 5.5 at $4 and $20, so the introductory price is half and the later price is equal. Wccftech, citing Artificial Analysis, puts the cost of one index task at $7.76 for Carbon, below Opus 5.5 (max) at $10.04 and above GPT-6 Astra (max) at $6.25.

Access starts with participants in Google's Fairwind program for cyber defenders; no date is set for developers or consumers. VentureBeat reports Claude Code with Opus 5.5 scored 62 on the Coding Agent Index, ahead of Carbon's 59, and Opus led OSWorld 87.2% to 81.5%. Scores Google published in its own table, including security results, have not been independently checked.