From Hallucination to Grounding: leaderboard
Current VLM evaluations often conflate language priors with genuine spatial reasoning. To address this, we introduce CRISP, a novel structural-diagnostic evaluation paradigm that assesses visual spatial intelligence through consistency, the alignment between implicit perception and explicit reasoning. Unlike traditional black-box QA, CRISP utilizes metric 3D Scene Graphs and an oracle intervention
Metric: Con. (self-reported). Source: benchmarklist.com. Saturation forecast: Around April 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 57.28 | #152 |
| 2 | GPT-5 Mini | 56.75 | #186 |
| 3 | Gemini 3 Flash | 55.2 | #99 |
| 4 | Gemini 2.5 Flash | 49.98 | #245 |
| 5 | GPT-5.2 | 48.87 | #110 |
| 6 | Qwen 2.5 VL 7B | 35.62 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=from-hallucination-to-grounding · How It Works · Data refreshed daily, snapshot 2026-10-20.