# Multi-view spatial reasoning on MindCube Source: [Multi-view spatial reasoning on MindCube]() Collection index: [llms.txt]() Data updated: 2026-06-12T08:29:08.479Z Evaluation of model performance on the MindCube dataset, specifically testing Multi-view spatial reasoning capabilities. 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 | Multi-view spatial reasoning | | Dataset (including any stated split) | MindCube | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 37 of 37 | | 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 - [Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | ViewFusion | 77 | Not recorded | 69ae53dd2d603e6de3e42872 | | SpatialClaw | 72.8 | Not recorded | 6a2b92d8d33dfb23e783b328 | | Gemini-3-Pro-Preview | 70.8 | Not recorded | 69ae53dd2d603e6de3e42872 | | ViewFusion | 68.5 | Not recorded | 69ae53dd2d603e6de3e42872 | | pySpatial | 67.1 | Not recorded | 6a2b92d8d33dfb23e783b328 | | Gemini-2.5-Pro | 57.6 | Not recorded | 69ae53dd2d603e6de3e42872 | | GPT-5 | 56.3 | Not recorded | 69ae53dd2d603e6de3e42872 | | SpaceTools Toolshed | 52.9 | Not recorded | 6a2b92d8d33dfb23e783b328 | | SpatialLadder-3B | 43.4 | Not recorded | 69ae53dd2d603e6de3e42872 | | InternVL3-8B | 41.5 | Not recorded | 69ae53dd2d603e6de3e42872 | | No-Think | 41.1 | Not recorded | 6a167ad9dbd7150506f802fe | | Cambrian-S-7B | 39.6 | Not recorded | 69ae53dd2d603e6de3e42872 | | ThinkMorph (24K) | 39.2 | Not recorded | 6a167ad9dbd7150506f802fe | | VST-7B-RL | 39.1 | Not recorded | 69ae53dd2d603e6de3e42872 | | SpaceR-7B | 37.9 | Not recorded | 69ae53dd2d603e6de3e42872 | | Qwen2.5-VL-3B-Instruct | 37.6 | Not recorded | 69ae53dd2d603e6de3e42872 | | InternVL3-2B | 37.5 | Not recorded | 69ae53dd2d603e6de3e42872 | | Qwen3-VL-4B-Instruct | 37 | Not recorded | 69ae53dd2d603e6de3e42872 | | Panoramic | 36.9 | Not recorded | 6a167ad9dbd7150506f802fe | | Top-down + VDrop | 36.5 | Not recorded | 6a167ad9dbd7150506f802fe | | VST-3B-RL | 36.4 | Not recorded | 69ae53dd2d603e6de3e42872 | | Qwen2.5-VL-7B-Instruct | 36 | Not recorded | 69ae53dd2d603e6de3e42872 | | Point Matching | 35.2 | Not recorded | 6a167ad9dbd7150506f802fe | | Top-down | 35.2 | Not recorded | 6a167ad9dbd7150506f802fe | | ViLaSR-7B | 35.1 | Not recorded | 69ae53dd2d603e6de3e42872 | | Qwen3-VL-2B-Instruct | 34.5 | Not recorded | 69ae53dd2d603e6de3e42872 | | Qwen3-VL-8B | 34.4 | Not recorded | 6a167ad9dbd7150506f802fe | | Panoramic + VDrop | 34.1 | Not recorded | 6a167ad9dbd7150506f802fe | | Point Matching + VDrop | 34.1 | Not recorded | 6a167ad9dbd7150506f802fe | | Spatial-MLLM-4B | 33.4 | Not recorded | 69ae53dd2d603e6de3e42872 | | RandomChoice | 33 | Not recorded | 69ae53dd2d603e6de3e42872 | | Cambrian-S-3B | 32.5 | Not recorded | 69ae53dd2d603e6de3e42872 | | BAGEL | 31.7 | Not recorded | 6a167ad9dbd7150506f802fe | | Qwen3-VL-8B-Instruct | 29.4 | Not recorded | 69ae53dd2d603e6de3e42872 | | Qwen3-VL-4B | 29 | Not recorded | 6a167ad9dbd7150506f802fe | | Text CoT | 25.3 | Not recorded | 6a167ad9dbd7150506f802fe | | BAGEL-Zebra-CoT (182K) | 21.7 | Not recorded | 6a167ad9dbd7150506f802fe | ### ViewFusion Accuracy: 77. Result ID: 69ae7e2e6dbb7075b95671fd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"4B","Training Protocol":"SFT + RL"}. This record does not assert independent reproduction. ### SpatialClaw Accuracy: 72.8. Result ID: 6a2bc33efbc373592c2a3a12. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Gemma4-31B"}. This record does not assert independent reproduction. ### Gemini-3-Pro-Preview Accuracy: 70.8. Result ID: 69ae7e2e6dbb7075b95671ec. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"-"}. This record does not assert independent reproduction. ### ViewFusion Accuracy: 68.5. Result ID: 69ae7e2e6dbb7075b95671fc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"4B","Training Protocol":"SFT"}. This record does not assert independent reproduction. ### pySpatial Accuracy: 67.1. Result ID: 6a2bc33efbc373592c2a3a10. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Gemma4-31B"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Accuracy: 57.6. Result ID: 69ae7e2e6dbb7075b95671ea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"-"}. This record does not assert independent reproduction. ### GPT-5 Accuracy: 56.3. Result ID: 69ae7e2e6dbb7075b95671eb. [Reporting paper]() Recorded evaluation configuration: {"Size":"-"}. This record does not assert independent reproduction. ### SpaceTools Toolshed Accuracy: 52.9. Result ID: 6a2bc33efbc373592c2a3a11. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Gemma4-31B"}. This record does not assert independent reproduction. ### SpatialLadder-3B Accuracy: 43.4. Result ID: 69ae7e2e6dbb7075b95671f4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3B"}. This record does not assert independent reproduction. ### InternVL3-8B Accuracy: 41.5. Result ID: 69ae7e2e6dbb7075b95671ee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B"}. This record does not assert independent reproduction. ### No-Think Accuracy: 41.1. Result ID: 6a16c560fbc373592c108b57. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Cambrian-S-7B Accuracy: 39.6. Result ID: 69ae7e2e6dbb7075b95671f9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction. ### ThinkMorph (24K) Accuracy: 39.2. Result ID: 6a16c560fbc373592c108b56. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### VST-7B-RL Accuracy: 39.1. Result ID: 69ae7e2e6dbb7075b95671fb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"7B","Training Protocol":"RL"}. This record does not assert independent reproduction. ### SpaceR-7B Accuracy: 37.9. Result ID: 69ae7e2e6dbb7075b95671f6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-Instruct Accuracy: 37.6. Result ID: 69ae7e2e6dbb7075b95671ef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3B","Training Protocol":"Instruct"}. This record does not assert independent reproduction. ### InternVL3-2B Accuracy: 37.5. Result ID: 69ae7e2e6dbb7075b95671ed. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"2B"}. This record does not assert independent reproduction. ### Qwen3-VL-4B-Instruct Accuracy: 37. Result ID: 69ae7e2e6dbb7075b95671f2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"4B","Training Protocol":"Instruct"}. This record does not assert independent reproduction. ### Panoramic Accuracy: 36.9. Result ID: 6a16c560fbc373592c108b59. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Top-down + VDrop Accuracy: 36.5. Result ID: 6a16c560fbc373592c108b5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✓"}. This record does not assert independent reproduction. ### VST-3B-RL Accuracy: 36.4. Result ID: 69ae7e2e6dbb7075b95671fa. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"3B","Training Protocol":"RL"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B-Instruct Accuracy: 36. Result ID: 69ae7e2e6dbb7075b95671f0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B","Training Protocol":"Instruct"}. This record does not assert independent reproduction. ### Point Matching Accuracy: 35.2. Result ID: 6a16c560fbc373592c108b5b. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Top-down Accuracy: 35.2. Result ID: 6a16c560fbc373592c108b5d. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### ViLaSR-7B Accuracy: 35.1. Result ID: 69ae7e2e6dbb7075b95671f7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B"}. This record does not assert independent reproduction. ### Qwen3-VL-2B-Instruct Accuracy: 34.5. Result ID: 69ae7e2e6dbb7075b95671f1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"2B","Training Protocol":"Instruct"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Accuracy: 34.4. Result ID: 6a16c560fbc373592c108b53. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Panoramic + VDrop Accuracy: 34.1. Result ID: 6a16c560fbc373592c108b5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✓"}. This record does not assert independent reproduction. ### Point Matching + VDrop Accuracy: 34.1. Result ID: 6a16c560fbc373592c108b5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✓"}. This record does not assert independent reproduction. ### Spatial-MLLM-4B Accuracy: 33.4. Result ID: 69ae7e2e6dbb7075b95671f5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"4B"}. This record does not assert independent reproduction. ### RandomChoice Accuracy: 33. Result ID: 69ae7e2e6dbb7075b95671e9. [Reporting paper]() Recorded evaluation configuration: {"Size":"-"}. This record does not assert independent reproduction. ### Cambrian-S-3B Accuracy: 32.5. Result ID: 69ae7e2e6dbb7075b95671f8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"3B"}. This record does not assert independent reproduction. ### BAGEL Accuracy: 31.7. Result ID: 6a16c560fbc373592c108b54. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Instruct Accuracy: 29.4. Result ID: 69ae7e2e6dbb7075b95671f3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Training Protocol":"Instruct"}. This record does not assert independent reproduction. ### Qwen3-VL-4B Accuracy: 29. Result ID: 6a16c560fbc373592c108b52. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### Text CoT Accuracy: 25.3. Result ID: 6a16c560fbc373592c108b58. [Reporting paper]() Recorded evaluation configuration: {"VDrop":"✗"}. This record does not assert independent reproduction. ### BAGEL-Zebra-CoT (182K) Accuracy: 21.7. Result ID: 6a16c560fbc373592c108b55. [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.