# Long-Video Understanding on LVBench 30m~2h Source: [Long-Video Understanding on LVBench 30m~2h]() Collection index: [llms.txt]() Data updated: 2026-03-04T09:24:21.928Z Evaluation of Long-Video Understanding capabilities using the 30m to 2h duration subset of the LVBench dataset, measured by 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 | Long-Video Understanding | | Dataset (including any stated split) | LVBench 30m~2h | | Metric | Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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 - [Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Score | Result date | Report paper | | --- | --- | --- | --- | | Qwen2.5+FlexSelect | 56.6 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | Nemotron-Nano-V2 VL + All layer reduction | 54.29 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | LLaVA-Video+FlexSelect | 52.9 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | Qwen2.5 VL+FlexSelect | 51.2 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | Qwen3 VL + All layer reduction | 45.45 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | VideoLLaMA3 | 45.3 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | SF-LLaVA-1.5 | 45.3 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | ViLAMP | 45.2 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | Qwen2-VL | 42.4 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | GPT-4o | 34.7 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | Gemini-1.5-Pro | 33.1 | Not recorded | 69a66bdf2d603e6de3e3ccfd | | Oryx-1.5 | 30.8 | Not recorded | 69a66bdf2d603e6de3e3ccfd | ### Qwen2.5+FlexSelect Score: 56.6. Result ID: 69a7f8e86dbb7075b94d112f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"72B"}. This record does not assert independent reproduction. ### Nemotron-Nano-V2 VL + All layer reduction Score: 54.29. Result ID: 69a7f8e86dbb7075b94d1131. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Hybrid","Size":"12B"}. This record does not assert independent reproduction. ### LLaVA-Video+FlexSelect Score: 52.9. Result ID: 69a7f8e86dbb7075b94d112e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"7B"}. This record does not assert independent reproduction. ### Qwen2.5 VL+FlexSelect Score: 51.2. Result ID: 69a7f8e86dbb7075b94d112d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"7B"}. This record does not assert independent reproduction. ### Qwen3 VL + All layer reduction Score: 45.45. Result ID: 69a7f8e86dbb7075b94d1130. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"8B"}. This record does not assert independent reproduction. ### VideoLLaMA3 Score: 45.3. Result ID: 69a7f8e86dbb7075b94d1129. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"7B"}. This record does not assert independent reproduction. ### SF-LLaVA-1.5 Score: 45.3. Result ID: 69a7f8e86dbb7075b94d112b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"7B"}. This record does not assert independent reproduction. ### ViLAMP Score: 45.2. Result ID: 69a7f8e86dbb7075b94d112c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"7B"}. This record does not assert independent reproduction. ### Qwen2-VL Score: 42.4. Result ID: 69a7f8e86dbb7075b94d1128. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"7B"}. This record does not assert independent reproduction. ### GPT-4o Score: 34.7. Result ID: 69a7f8e86dbb7075b94d1126. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"–"}. This record does not assert independent reproduction. ### Gemini-1.5-Pro Score: 33.1. Result ID: 69a7f8e86dbb7075b94d1127. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"–"}. This record does not assert independent reproduction. ### Oryx-1.5 Score: 30.8. Result ID: 69a7f8e86dbb7075b94d112a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Transformer","Size":"34B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.