SpaceDG: leaderboard
Multimodal Large Language Models (MLLMs) have made rapid progress in spatial intelligence, yet existing spatial reasoning benchmarks largely assume pristine visual inputs and overlook the degradations that commonly occur in real-world deployment, such as motion blur, low light, adverse weather, lens distortion, and compression artifacts. This raises a fundamental question: how robust is the spatia
Metric: Avg. (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Median Human | 59.5 | |
| 2 | Gemini 3.1 Pro (Preview) | 56.7 | #58 |
| 3 | Gemini 3.1 Flash Lite | 48.8 | #191 |
| 4 | GPT-5.4 | 46.2 | #79 |
| 5 | Qwen 3.6 Plus | 46 | #120 |
| 6 | Qwen 3 VL 32B Instruct | 45.2 | #283 |
| 7 | Claude Sonnet 4.6 | 44.2 | #90 |
| 8 | Qwen 3.5 27B | 43.8 | #187 |
| 9 | Qwen 3.6 35B A3B | 42.4 | #226 |
| 10 | Qwen 3 VL 8B Instruct | 42.1 | #420 |
| 11 | Qwen 3.5 9B | 41.7 | #379 |
| 12 | Qwen 3.5 4B | 40.4 | #475 |
| 13 | InternVL3-8B | 40.1 | #620 |
| 14 | Qwen 3 VL 4B Instruct | 39.7 | #515 |
| 15 | Grok 4.1 Fast | 35.6 | #217 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=spacedg · How It Works · Data refreshed daily, snapshot 2026-10-20.