Artificial Analysis Intelligence Index artificial-analysis-index Leaderboard
The Artificial Analysis Intelligence Index — the field's most-cited cross-model aggregate, a composite of GPQA Diamond, MMLU-Pro, AIME, LiveCodeBench, HLE and more across frontier closed + open-weight models. Higher is better. · Metric: Intelligence Index (higher is better) · 🟢 Updated 17h ago
| # | Model | Intelligence Index | Paper |
|---|---|---|---|
| 1 | Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | 59.90 | — |
| 2 | GPT-5.6 Sol (max) | 58.90 | — |
| 3 | GPT-5.6 Sol (xhigh) | 57.70 | — |
| 4 | Kimi K3 | 57.10 | — |
| 5 | GPT-5.6 Sol (high) | 55.90 | — |
| 6 | Claude Opus 4.8 (Adaptive Reasoning, Max Effort) | 55.70 | — |
| 7 | GPT-5.6 Terra (max) | 55.00 | — |
| 8 | GPT-5.5 (xhigh) | 54.80 | — |
| 9 | Grok 4.5 (high) | 53.80 | — |
| 10 | GPT-5.6 Sol (medium) | 53.60 | — |
| 11 | Claude Opus 4.7 (Adaptive Reasoning, Max Effort) | 53.50 | — |
| 12 | Claude Sonnet 5 (Adaptive Reasoning, Max Effort) | 53.40 | — |
| 13 | GPT-5.5 (high) | 53.10 | — |
| 14 | GPT-5.6 Terra (xhigh) | 51.60 | — |
| 15 | GPT-5.4 (xhigh) | 51.40 | — |
| 16 | GPT-5.6 Luna (max) | 51.20 | — |
| 17 | GLM-5.2 (max) | 51.10 | — |
| 18 | Muse Spark 1.1 (xhigh) | 50.60 | — |
| 19 | GPT-5.5 (medium) | 50.40 | — |
| 20 | Gemini 3.5 Flash (high) | 50.20 | — |
| 21 | GPT-5.6 Sol (low) | 49.40 | — |
| 22 | GPT-5.6 Luna (xhigh) | 49.10 | — |
| 23 | GPT-5.6 Terra (high) | 49.00 | — |
| 24 | Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) | 47.20 | — |
| 25 | Gemini 3.1 Pro Preview | 46.50 | — |
| 26 | GPT-5.6 Luna (high) | 46.10 | — |
| 27 | Qwen3.7 Max | 46.00 | — |
| 28 | GPT-5.6 Terra (medium) | 45.60 | — |
| 29 | Gemini 3.5 Flash (medium) | 45.40 | — |
| 30 | MiniMax-M3 | 44.40 | — |