Google's Gemini 4 Carbon has taken the #1 spot on the Artificial Analysis Intelligence Index v5, scoring 59 and leading in complex reasoning, knowledge and scientific tasks. It ranks first in 4 of the 9 Index evaluations, including HLE-Diamond, Terminal-Bench-Science, AA-Omniscience and Automation Bench-AA.
In coding, Gemini 4 Carbon scores 59 on the Artificial Analysis Coding Agent Index with Antigravity CLI, just 3 points behind Claude Code with Claude Opus 5.5 (max, 62). It ranks second on both Terminal-Bench 4.0 (64%) and Frontier Code v1.1 (51%), behind Claude Opus 5.5. Google has historically underperformed in this area, so the improvement stands out.
In agentic knowledge work, Gemini 4 Carbon scores 1609 Elo on AA-Briefcase, a 126-point gain over Gemini 4 Argon. It still trails models including Grok 4.7, Muse Wonder and Kimi K3.1 Pro. The largest gain came in Presentation, up 293 Elo to 1598. On AA-Omniscience v1.1, it scores 54, six points ahead of Claude Opus 5.5 (max), with accuracy rising to 58% and a hallucination rate of 11%, second lowest of any model tested.
Cost is also competitive. At standard pricing, Gemini 4 Carbon costs $7.76 per Intelligence Index task, about 23% less than Claude Opus 5.5 (max, $10.04), and uses 29% fewer output tokens. The model offers a 1M-token context window and accepts text, image, video and speech input. It is priced at $4/$20 per 1M input/output tokens, with a 50% launch discount that has no confirmed end date. It is currently rolling out to selected users and is not yet publicly available.
Source: Hacker News · Summarized by HeadlinesBriefing