# Open-ended Video Question Answering on NExT-QA (val) Source: [Open-ended Video Question Answering on NExT-QA (val)]() Collection index: [llms.txt]() Data updated: 2026-06-12T08:09:53.117Z Evaluation of model performance on the open-ended subset of the NExT-QA validation split for Video Question Answering, reporting WUPS score. 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 | Open-ended Video Question Answering | | Dataset (including any stated split) | NExT-QA open-ended (val) | | Metric | WUPS | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 26 of 26 | | 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 - [WUPS](): Higher is better; unit: PERCENT. ## Recorded results | Method | WUPS | Result date | Report paper | | --- | --- | --- | --- | | InternVL3 + ReFoCUS | 26.8 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3 | 26.6 | Not recorded | 68fa79001e209131cea7b967 | | VideoLLaMA 3 + ReFoCUS | 26.5 | Not recorded | 68fa79001e209131cea7b967 | | VideoLLaMA 3 | 25.8 | Not recorded | 68fa79001e209131cea7b967 | | Qwen3-VL + ReFoCUS | 25.7 | Not recorded | 68fa79001e209131cea7b967 | | Qwen3-VL | 25.3 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3 + ReFoCUS | 25 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3.5 + ReFoCUS | 24.7 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3 | 24.4 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3.5 | 24.3 | Not recorded | 68fa79001e209131cea7b967 | | Qwen3-VL + ReFoCUS | 24.1 | Not recorded | 68fa79001e209131cea7b967 | | Qwen3-VL | 23.8 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3.5 + ReFoCUS | 22.9 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3.5 | 22.1 | Not recorded | 68fa79001e209131cea7b967 | | VideoLLaMA 3 + ReFoCUS | 20.7 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3 + ReFoCUS | 20.4 | Not recorded | 68fa79001e209131cea7b967 | | InternVL3 | 20 | Not recorded | 68fa79001e209131cea7b967 | | LLaVA-OneVision + ReFoCUS | 18.9 | Not recorded | 68fa79001e209131cea7b967 | | VideoLLaMA 3 | 18.9 | Not recorded | 68fa79001e209131cea7b967 | | LLaVA-OneVision | 18.1 | Not recorded | 68fa79001e209131cea7b967 | | LLaVA-OneVision + ReFoCUS | 16.4 | Not recorded | 68fa79001e209131cea7b967 | | LLaVA-OneVision | 16.2 | Not recorded | 68fa79001e209131cea7b967 | | Gemini 2.5 Flash + ReFoCUS | 11.9 | Not recorded | 68fa79001e209131cea7b967 | | Gemini 2.5 Flash | 11.7 | Not recorded | 68fa79001e209131cea7b967 | | GPT-4o + ReFoCUS | 9.1 | Not recorded | 68fa79001e209131cea7b967 | | GPT-4o | 8.5 | Not recorded | 68fa79001e209131cea7b967 | ### InternVL3 + ReFoCUS WUPS: 26.8. Result ID: 6a2bbebcfbc373592c2a1783. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"8B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3 WUPS: 26.6. Result ID: 6a2bbebcfbc373592c2a1782. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"8B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### VideoLLaMA 3 + ReFoCUS WUPS: 26.5. Result ID: 6a2bbebcfbc373592c2a177f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"7B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### VideoLLaMA 3 WUPS: 25.8. Result ID: 6a2bbebcfbc373592c2a177e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"7B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### Qwen3-VL + ReFoCUS WUPS: 25.7. Result ID: 6a2bbebcfbc373592c2a1787. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"8B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### Qwen3-VL WUPS: 25.3. Result ID: 6a2bbebcfbc373592c2a1786. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"8B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3 + ReFoCUS WUPS: 25. Result ID: 6a2bbebcfbc373592c2a1779. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"2B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3.5 + ReFoCUS WUPS: 24.7. Result ID: 6a2bbebcfbc373592c2a1785. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"8B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3 WUPS: 24.4. Result ID: 6a2bbebcfbc373592c2a1778. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"2B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3.5 WUPS: 24.3. Result ID: 6a2bbebcfbc373592c2a1784. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"8B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### Qwen3-VL + ReFoCUS WUPS: 24.1. Result ID: 6a2bbebcfbc373592c2a177d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"4B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### Qwen3-VL WUPS: 23.8. Result ID: 6a2bbebcfbc373592c2a177c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"4B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3.5 + ReFoCUS WUPS: 22.9. Result ID: 6a2bbebcfbc373592c2a177b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"4B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3.5 WUPS: 22.1. Result ID: 6a2bbebcfbc373592c2a177a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"4B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### VideoLLaMA 3 + ReFoCUS WUPS: 20.7. Result ID: 6a2bbebcfbc373592c2a1777. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"2B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3 + ReFoCUS WUPS: 20.4. Result ID: 6a2bbebcfbc373592c2a1775. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"1B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### InternVL3 WUPS: 20. Result ID: 6a2bbebcfbc373592c2a1774. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"1B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### LLaVA-OneVision + ReFoCUS WUPS: 18.9. Result ID: 6a2bbebcfbc373592c2a1773. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"0.5B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### VideoLLaMA 3 WUPS: 18.9. Result ID: 6a2bbebcfbc373592c2a1776. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"2B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### LLaVA-OneVision WUPS: 18.1. Result ID: 6a2bbebcfbc373592c2a1772. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM Size":"0.5B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### LLaVA-OneVision + ReFoCUS WUPS: 16.4. Result ID: 6a2bbebcfbc373592c2a1781. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"7B","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### LLaVA-OneVision WUPS: 16.2. Result ID: 6a2bbebcfbc373592c2a1780. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM Size":"7B","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### Gemini 2.5 Flash + ReFoCUS WUPS: 11.9. Result ID: 6a2bbebcfbc373592c2a176f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"Closed","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### Gemini 2.5 Flash WUPS: 11.7. Result ID: 6a2bbebcfbc373592c2a176e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"Closed","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### GPT-4o + ReFoCUS WUPS: 9.1. Result ID: 6a2bbebcfbc373592c2a1771. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Size":"Closed","Sampling Strategy":"ReFoCUS","Frame Budget":"32-frame"}. This record does not assert independent reproduction. ### GPT-4o WUPS: 8.5. Result ID: 6a2bbebcfbc373592c2a1770. [Reporting paper]() Recorded evaluation configuration: {"LLM Size":"Closed","Sampling Strategy":"Uniform sampling","Frame Budget":"32-frame"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.