CAIS Text Capabilities Index: leaderboard
Composite CAIS AI Dashboard text index averaging Humanity's Last Exam, ARC-AGI-2, TextQuests, and SWE-bench Pro for models with all component scores.
Metric: Text Capabilities Index (self-reported). Source: benchmarklist.com. Status: saturation imminent. 40 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 55.9 |
| 2 | Claude Opus 4.8 | 53.8 |
| 3 | Gemini 3.1 Pro (Preview) | 52.9 |
| 4 | GPT-5.4 | 49.3 |
| 5 | Gemini 3.5 Flash | 48.9 |
| 6 | Claude Opus 4.7 | 46.9 |
| 7 | Claude Opus 4.6 | 44 |
| 8 | Gemini 3 Pro | 38.4 |
| 9 | Claude Opus 4.5 | 36.6 |
| 10 | Gemini 3 Flash | 35.6 |
| 11 | GPT-5.2 | 33.8 |
| 12 | Kimi K2.6 | 32.8 |
| 13 | Claude Sonnet 4.6 | 32.6 |
| 14 | Grok 4.20 | 32.5 |
| 15 | DeepSeek V4 Pro | 32.1 |
Interactive version: theaggregate.ai/benchmark?slug=cais-text-capabilities-index · How It Works · Data refreshed daily, snapshot 2026-09-05.