Artificial Analysis Intelligence Index
IntelligenceThe most-cited composite intelligence score: a 0–100 index combining knowledge, reasoning, math, coding, and agentic evaluations (GPQA Diamond, HLE, IFBench, SciCode, Terminal-Bench Hard, τ²-Bench, and more). Higher is better.
Leaderboard
#ModelScorePrice
- 1Claude Fable 5Adaptive Reasoning, Max Effort, Opus 4.8 Fallback64.9$21.88
- 2Claude Opus 4.8Adaptive Reasoning, Max Effort61.4$10.94
- 3GPT-5.5xhigh60.2$11.25
- 4GPT-5.5high58.9$11.25
- 5Claude Opus 4.7Adaptive Reasoning, Max Effort57.3$10.94
- 657.2$4.50
- 7GPT-5.5medium56.7$11.25
- 856.6$3.75
- 9Gemini 3.5 Flashhigh55.3$3.38
- 10Gemini 3.5 Flashmedium54.8$3.38
- 1154.7$0.53
- 1253.9$1.71
- 13Grok 4.3high53.2$1.56
- 14Claude Opus 4.7Non-reasoning, High Effort51.8$10.94
- 15Claude Sonnet 4.6Adaptive Reasoning, Max Effort51.7$6.56
- 16DeepSeek V4Reasoning, Max Effort51.5$0.54
- 17GPT-5.2xhigh51.3$4.81
- 18GPT-5.5low50.8$11.25
- 19DeepSeek V4Reasoning, High Effort49.8$0.54
- 20GPT-5.2 Codexxhigh49$4.81
- 21Grok 4.3medium48.8$1.56
- 22Gemini 3 Prohigh48.4$4.50
- 23Kimi K2.5Reasoning46.8$1.19
- 24GPT-5.2medium46.6$4.81
- 25Gemini 3 FlashReasoning46.4$1.13
- 26Claude Sonnet 4.6Non-reasoning, High Effort44.4$6.56
- 27Grok 4.3low43.9$1.56
- 28Gemini 3.5 Flashminimal43.3$3.38
- 29Kimi K2.6Non-reasoning42.9$1.71
- 30Claude Sonnet 4.6Non-reasoning, Low Effort42.6$6.56
- 31GPT-5.5May 202641.8$11.25
- 32Gemini 3 Prolow41.3$4.50
- 33GPT-5.5Non-reasoning40.9$11.25
- 34DeepSeek V4Non-reasoning39.3$0.54
- 35Kimi K2.5Non-reasoning37.3$1.20
- 36Claude Haiku 4.5Reasoning37.1$2.19
- 37Gemini 3 FlashNon-reasoning35$1.13
- 38GPT-5.2Non-reasoning33.6$4.81
- 39Claude Haiku 4.5Non-reasoning31$2.19
- 40Grok 4.3Non-reasoning31$1.56
- 4122.8$0.75
Details:
Category
Intelligence- Created by
Artificial Analysis Models tested
19Configs tested
41Leader
Claude Fable 5Top score
64.9
Updated June 2026