Gemini 4 Carbon tops Artificial Analysis Intelligence Index v5

In coding agent tasks, Gemini 4 Carbon ranks second only to Claude Opus 5.5. It scores 59 on the Artificial Analysis Coding Agent Index, just three points behind Claude Opus 5.5’s score of 62. This marks a significant improvement for Google, which has historically lagged in this specific area. The model also shows gains in agentic knowledge work, improving its Elo score by 126 points over the previous Gemini 4 Argon, though it remains behind models like Grok 4.7 and Kimi K3.1 Pro.

At standard pricing of $4 per million input tokens and $20 per million output tokens, Gemini 4 Carbon costs approximately $7.76 per Intelligence Index task. This is about 23% less than Claude Opus 5.5 and slightly higher than GPT-6 Astra. The model supports a 1M token context window and accepts text, image, video, and speech inputs, though it only outputs text. Its hallucination rate is 11%, which is lower than competitors like Claude Opus 5.5 and GPT-6 Astra, both at 50%.