Quick Take
  • Google’s Gemini 4 Carbon took first place on the Artificial Analysis Intelligence Index, an independent ranking of AI models, on Wednesday.
  • Carbon scored 59, edging Anthropic’s Claude Opus 5.5 at 58 and OpenAI’s GPT-6 Astra at 57.
  • Alphabet (GOOGL) shares rose about 1.5% on the day.
  • Google is giving Carbon first to vetted cyber defenders through its Fairwind Program.

What Happened

Google said the staged release puts safety first. Carbon’s biggest lead sits on hacking-related tests.

On ExploitGym, which checks whether a model can prove a software flaw is usable in an attack, Carbon scored 54.8%. GPT-6 Astra scored 42.4% and Opus 5.5 scored 34.5%, according to Google’s benchmark table.

Carbon launches at $2 per million input tokens and $10 per million output tokens. Those prices later double to $4 and $20.

Market Context

OpenAI, by contrast, cut GPT-6.1 Terra prices to $1 and $5 on October 26. Anthropic, the maker of Claude, is preparing for its planned stock listing.

Google has not said when wider access opens, or when the higher prices begin. Meanwhile, all three labs have released new models this month.

Why It Matters

Google’s Gemini 4 Carbon took first place on the Artificial Analysis Intelligence Index, an independent ranking of AI models, on Wednesday. Yet most paying customers cannot buy it.

Carbon scored 59, edging Anthropic’s Claude Opus 5.5 at 58 and OpenAI’s GPT-6 Astra at 57. Alphabet (GOOGL) shares rose about 1.5% on the day.

Details

Why Can’t Most Customers Buy Gemini 4 Carbon?

Google is giving Carbon first to vetted cyber defenders through its Fairwind Program. Developers, businesses, and consumers will follow, but Google gave no date.

Google ran those outside cyber tests with safety filters switched off. It is also joining the US government’s voluntary pre-release model testing.

Google took the same path with Gemini 4 Argon, its previous top model. Argon opened to paying developers on October 13, roughly two weeks before Carbon arrived.

How Much Will Gemini 4 Carbon Cost?

Google charges for models like Carbon per token, the chunks of text a model reads and writes.

Carbon’s lead is also narrow. Most head-to-head figures come from Google’s own table, and the Artificial Analysis public leaderboard did not yet list Carbon when checked on Wednesday.

Rivals still win some tests. On FrontierSWE 2, a coding test, Astra scored 65.5% and Opus 62.3%, against Carbon’s 61.2%.

The post Gemini 4 Carbon Beats Claude and GPT-6: Why Is Google Holding It Back? appeared first on BeInCrypto.