AI Briefing
KOSign in

Gemini 4 Carbon Tops Artificial Analysis Intelligence Index v5

·2026.10.28 09:00

Key point

Google's Gemini 4 Carbon scores 59 on the Intelligence Index v5, leading in complex reasoning and scientific tasks while ranking second only to Claude Opus 5.5 on coding agents.

1 / 7

Details

  • Gemini 4 Carbon achieves a score of 59 on the Artificial Analysis Intelligence Index v5, ranking #1 in 4 of 9 evaluations including HLE-Diamond, Terminal-Bench-Science, AA-Omniscience, and AutomationBench-AA.
  • On the Coding Agent Index, Gemini 4 Carbon (high) scores 59 with Antigravity CLI, placing it 3 points behind Claude Opus 5.5 (max) at 62; it ranks second on Terminal-Bench 4.0 (64%) and FrontierCode v1.1 (51%).
  • The model costs $7.76 per Intelligence Index task, approximately 23% less than Claude Opus 5.5 ($10.04), using 123k output tokens per task compared to 173k for Claude Opus 5.5.
  • Gemini 4 Carbon is currently rolling out to selected users with a 50% launch discount, featuring a 1M token context window and multimodal input support for text, image, video, and speech.

Performance Details

  • AA-Omniscience v1.1 score is 54, leading Claude Opus 5.5 (48) and Gemini 4 Argon (46), with a hallucination rate of 11%.
  • Agentic knowledge work on AA-Briefcase shows a 126-point improvement over Gemini 4 Argon to 1609 Elo, though it remains behind Grok 4.7, Muse Wonder, and Kimi K3.1 Pro.
  • Standard pricing is $4 per 1M input tokens and $20 per 1M output tokens, with cached input tokens receiving a 95% discount.
googlebenchmarkartificial-analysisllm-evaluationcoding-agentsgemini-4-carbon
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

GeekNewsKimi K3.1 Pro Scores 55 on Artificial Analysis Intelligence IndexRedditArtificial Analysis Intelligence Index v5 Reveals 13x Cost Variance Across Frontier ModelsHacker NewsArtificial Analysis Intelligence Index v5: Claude Opus 5.5 Leads as Open Weights Gap Widens to 12 PointsRedditMeta's 816B Muse Spark Open scores 39 on Artificial Analysis, trailing 180B Qwen3.8-Flash-NextHuggingFace BlogAi2 releases FairStress benchmark revealing asymmetric robustness in LLM decision-makingHacker NewsOpenAI releases GPT-6.1 Terra with 50% lower API costs and improved computer use