# Visual Spatial Intelligence Reasoning on VSI-Debiased Source: [Visual Spatial Intelligence Reasoning on VSI-Debiased]() Collection index: [llms.txt]() Data updated: 2026-03-04T09:30:18.170Z Evaluation of model performance on the VSI-Debiased dataset for Visual Spatial Intelligence Reasoning, measured by Accuracy. 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 | Visual Spatial Intelligence Reasoning | | Dataset (including any stated split) | VSI-Debiased | | Metric | Accuracy | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 27 of 27 | | 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: SCALAR. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | SSR-3D | 0.699 | Not recorded | 69a66bc42d603e6de3e3caf5 | | GeoThinker Qwen3vl-8B-32frame | 0.681 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | GeoThinker Qwen3vl-8B-32frame | 0.677 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | SSR-2D | 0.666 | Not recorded | 69a66bc42d603e6de3e3caf5 | | GeoThinker Qwen3vl-8B-32frame | 0.663 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | GeoThinker Qwen3vl-8B-8frame | 0.653 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | GeoThinker Qwen3vl-8B-8frame | 0.648 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | GeoThinker Qwen3vl-8B-8frame | 0.643 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | GeoThinker Qwen3vl-8B-32frame | 0.643 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | SenseNova-SI InternVL3-8B | 0.628 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | SenseNova-SI (InternVL3-8B) | 0.628 | Not recorded | 69a66bc42d603e6de3e3caf5 | | SenseNova-SI InternVL3-8B | 0.624 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | GeoThinker Qwen3vl-8B-8frame | 0.607 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | Cambrian-S-7B | 0.599 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | SenseNova-SI InternVL3-8B | 0.597 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | Cambrian-S-7B | 0.591 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | SenseNova-SI InternVL3-8B | 0.589 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | Cambrian-S-7B | 0.563 | Not recorded | 69a66bc42d603e6de3e3caf5 | | Cambrian-S-7B | 0.556 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | VG-LLM-8B\* | 0.552 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | VG-LLM-8B\* | 0.551 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | VG-LLM-8B\* | 0.524 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | VG-LLM-8B\* | 0.516 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | Cambrian-S-7B | 0.497 | Not recorded | 698575e4c97a0b3e3b5d5f2a | | LLaVA-Video-7B | 0.307 | Not recorded | 69a66bc42d603e6de3e3caf5 | | LLaVA-OneVision-7B | 0.285 | Not recorded | 69a66bc42d603e6de3e3caf5 | | SmolVLM2-2.2B | 0.223 | Not recorded | 69a66bc42d603e6de3e3caf5 | ### SSR-3D Accuracy: 0.699. Result ID: 69a7fb9f6dbb7075b94d23a8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Open-Sourced spatial models"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-32frame Accuracy: 0.681. Result ID: 6997383f5698665ec4883e88. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"128"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-32frame Accuracy: 0.677. Result ID: 6997383f5698665ec4883e87. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"64"}. This record does not assert independent reproduction. ### SSR-2D Accuracy: 0.666. Result ID: 69a7fb9f6dbb7075b94d23a7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Open-Sourced spatial models"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-32frame Accuracy: 0.663. Result ID: 6997383f5698665ec4883e86. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"32"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-8frame Accuracy: 0.653. Result ID: 6997383f5698665ec4883e84. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"128"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-8frame Accuracy: 0.648. Result ID: 6997383f5698665ec4883e82. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"32"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-8frame Accuracy: 0.643. Result ID: 6997383f5698665ec4883e83. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"64"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-32frame Accuracy: 0.643. Result ID: 6997383f5698665ec4883e85. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"16"}. This record does not assert independent reproduction. ### SenseNova-SI InternVL3-8B Accuracy: 0.628. Result ID: 6997383f5698665ec4883e7e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"# Frames":"32"}. This record does not assert independent reproduction. ### SenseNova-SI (InternVL3-8B) Accuracy: 0.628. Result ID: 69a7fb9f6dbb7075b94d23a6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"Open-Sourced spatial models"}. This record does not assert independent reproduction. ### SenseNova-SI InternVL3-8B Accuracy: 0.624. Result ID: 6997383f5698665ec4883e7f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"# Frames":"64"}. This record does not assert independent reproduction. ### GeoThinker Qwen3vl-8B-8frame Accuracy: 0.607. Result ID: 6997383f5698665ec4883e81. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"16"}. This record does not assert independent reproduction. ### Cambrian-S-7B Accuracy: 0.599. Result ID: 6997383f5698665ec4883e78. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"128"}. This record does not assert independent reproduction. ### SenseNova-SI InternVL3-8B Accuracy: 0.597. Result ID: 6997383f5698665ec4883e80. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"# Frames":"128"}. This record does not assert independent reproduction. ### Cambrian-S-7B Accuracy: 0.591. Result ID: 6997383f5698665ec4883e77. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"64"}. This record does not assert independent reproduction. ### SenseNova-SI InternVL3-8B Accuracy: 0.589. Result ID: 6997383f5698665ec4883e7d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"# Frames":"16"}. This record does not assert independent reproduction. ### Cambrian-S-7B Accuracy: 0.563. Result ID: 69a7fb9f6dbb7075b94d23a5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Open-Sourced spatial models"}. This record does not assert independent reproduction. ### Cambrian-S-7B Accuracy: 0.556. Result ID: 6997383f5698665ec4883e76. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"32"}. This record does not assert independent reproduction. ### VG-LLM-8B\* Accuracy: 0.552. Result ID: 6997383f5698665ec4883e7b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"64"}. This record does not assert independent reproduction. ### VG-LLM-8B\* Accuracy: 0.551. Result ID: 6997383f5698665ec4883e7c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"128"}. This record does not assert independent reproduction. ### VG-LLM-8B\* Accuracy: 0.524. Result ID: 6997383f5698665ec4883e7a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"32"}. This record does not assert independent reproduction. ### VG-LLM-8B\* Accuracy: 0.516. Result ID: 6997383f5698665ec4883e79. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"16"}. This record does not assert independent reproduction. ### Cambrian-S-7B Accuracy: 0.497. Result ID: 6997383f5698665ec4883e75. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"# Frames":"16"}. This record does not assert independent reproduction. ### LLaVA-Video-7B Accuracy: 0.307. Result ID: 69a7fb9f6dbb7075b94d23a3. [Reporting paper]() Recorded evaluation configuration: {"Category":"Open-Sourced general models"}. This record does not assert independent reproduction. ### LLaVA-OneVision-7B Accuracy: 0.285. Result ID: 69a7fb9f6dbb7075b94d23a2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Category":"Open-Sourced general models"}. This record does not assert independent reproduction. ### SmolVLM2-2.2B Accuracy: 0.223. Result ID: 69a7fb9f6dbb7075b94d23a4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Category":"Open-Sourced general models"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.