# OOD Generalization on Real-world Cross-View Spatial Reasoning Benchmarks (Aggregate) Source: [OOD Generalization on Real-world Cross-View Spatial Reasoning Benchmarks (Aggregate)]() Collection index: [llms.txt]() Data updated: 2026-05-27T10:21:13.147Z Evaluation of model generalization capabilities across an aggregate suite of benchmarks focused on real-world, cross-view spatial reasoning tasks, reporting the average Out-of-Distribution (OOD) performance. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Generalization across real-world cross-view spatial reasoning benchmarks | | Dataset (including any stated split) | OOD Evaluation Suite Aggregate | | Metric | Avg OOD Performance | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Avg OOD Performance](): Higher is better; unit: PERCENT. ## Recorded results | Method | Avg OOD Performance | Result date | Report paper | | --- | --- | --- | --- | | Panoramic + VDrop | 40 | Not recorded | 6a167ad9dbd7150506f802fe | | Top-down + VDrop | 38 | Not recorded | 6a167ad9dbd7150506f802fe | | Panoramic | 37.6 | Not recorded | 6a167ad9dbd7150506f802fe | | Top-down | 37.3 | Not recorded | 6a167ad9dbd7150506f802fe | | ThinkMorph (24K) | 37.2 | Not recorded | 6a167ad9dbd7150506f802fe | | Qwen3-VL-8B | 37 | Not recorded | 6a167ad9dbd7150506f802fe | | Point Matching + VDrop | 36.1 | Not recorded | 6a167ad9dbd7150506f802fe | | Point Matching | 35.2 | Not recorded | 6a167ad9dbd7150506f802fe | | No-Think | 35.1 | Not recorded | 6a167ad9dbd7150506f802fe | | Qwen3-VL-4B | 33.8 | Not recorded | 6a167ad9dbd7150506f802fe | | BAGEL | 33.3 | Not recorded | 6a167ad9dbd7150506f802fe | | Text CoT | 32.7 | Not recorded | 6a167ad9dbd7150506f802fe | | BAGEL-Zebra-CoT (182K) | 26.8 | Not recorded | 6a167ad9dbd7150506f802fe | ### Panoramic + VDrop Avg OOD Performance: 40. Result ID: 6a16c560fbc373592c108b8e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✓"}. This record does not assert independent reproduction. ### Top-down + VDrop Avg OOD Performance: 38. Result ID: 6a16c560fbc373592c108b92. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✓"}. This record does not assert independent reproduction. ### Panoramic Avg OOD Performance: 37.6. Result ID: 6a16c560fbc373592c108b8d. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Top-down Avg OOD Performance: 37.3. Result ID: 6a16c560fbc373592c108b91. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### ThinkMorph (24K) Avg OOD Performance: 37.2. Result ID: 6a16c560fbc373592c108b8a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Avg OOD Performance: 37. Result ID: 6a16c560fbc373592c108b87. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Point Matching + VDrop Avg OOD Performance: 36.1. Result ID: 6a16c560fbc373592c108b90. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✓"}. This record does not assert independent reproduction. ### Point Matching Avg OOD Performance: 35.2. Result ID: 6a16c560fbc373592c108b8f. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### No-Think Avg OOD Performance: 35.1. Result ID: 6a16c560fbc373592c108b8b. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Qwen3-VL-4B Avg OOD Performance: 33.8. Result ID: 6a16c560fbc373592c108b86. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### BAGEL Avg OOD Performance: 33.3. Result ID: 6a16c560fbc373592c108b88. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Text CoT Avg OOD Performance: 32.7. Result ID: 6a16c560fbc373592c108b8c. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### BAGEL-Zebra-CoT (182K) Avg OOD Performance: 26.8. Result ID: 6a16c560fbc373592c108b89. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.