AI Decathlon - Spatial Reasoning: leaderboard
AI Decathlon Timed Challenge (Spatial Reasoning) evaluating coding agents on 60-100 minute engineering, math, spatial, and interactive design tasks.
Metric: Spatial Reasoning Score (0-100). Source: octathlon.wendystars.com. Saturation forecast: Around February 2027. 25 models tracked. Updated 22 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 4 Neon | xHigh | 99.4 | #3 |
| 2 | Gemini 4 Carbon | High | 80.5 | #7 |
| 3 | GPT-6.2 Astra | xHigh | 79.5 | #4 |
| 4 | GPT-6.2 Sol | xHigh | 77.9 | #10 |
| 5 | GPT-6.1 Sol | xHigh | 69.4 | #15 |
| 6 | GPT-6 Astra | xHigh | 66.7 | #11 |
| 7 | Claude Fable 5.2 | xHigh | 63.2 | #2 |
| 8 | Gemini 4 Hydrogen | xHigh | 58.3 | #16 |
| 9 | GPT-6.1 Terra | xHigh | 28.1 | #25 |
| 10 | Claude Opus 5.5 | xHigh | 28 | #9 |
| 11 | GPT-5.6 Sol | xHigh | 26.7 | #36 |
| 12 | Claude Sonnet 5.5 | xHigh | 26.5 | #24 |
| 13 | Grok 4.7 | xHigh | 18.6 | #55 |
| 14 | Qwen 4 Max (Preview) | xHigh | 10.4 | #63 |
| 15 | DeepSeek V4.1 Flash | Max | 8 | #69 |
Interactive version: theaggregate.ai/benchmark?slug=ai-decathlon-spatial-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-22.