TrackingAI IQ Test (Offline): leaderboard
A novel IQ test created by a Mensa member that has never appeared on the public internet, designed to avoid data contamination in AI training sets.
Metric: Offline IQ Score (%). Source: trackingai.org. Saturation forecast: Around January 2027. 40 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (Max) | 87.5 | #12 (Claude Opus 5) |
| 2 | Kimi K3 | 81.25 | #30 |
| 3 | Claude Opus 5 | 81.25 | #12 |
| 4 | Claude Fable 5 | 81.25 | #16 |
| 5 | GPT-5 Pro | 81.25 | #81 |
| 6 | GPT-5 | 75 | #98 |
| 7 | Grok 4 | 75 | #181 |
| 8 | Gemini 3 Pro | 75 | #80 |
| 9 | Claude Sonnet 5 | 75 | #57 |
| 10 | Claude Opus 4 (Thinking) | 75 | #163 (Claude Opus 4) |
| 11 | GPT-5 (Thinking) | 75 | #98 (GPT-5) |
| 12 | Claude Sonnet 4 | 68.75 | #199 |
| 13 | Grok 4.5 | 68.75 | #48 |
| 14 | O3 Pro | 68.75 | #116 |
| 15 | O3 | 62.5 | #124 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=trackingai-iq-test-offline · How It Works · Data refreshed daily, snapshot 2026-10-20.