# Video Understanding on Video-MME Long (Accuracy w/o/w Subtitles) Source: [Video Understanding on Video-MME Long (Accuracy w/o/w Subtitles)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:34:18.937Z Evaluation of general video understanding capabilities on the Video-MME Long dataset, specifically reporting accuracy metrics distinguishing performance with and without subtitles. 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 | Video Understanding | | Dataset (including any stated split) | Video-MME Long | | Metric | Accuracy (Long, wo Sub) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 32 of 32 | | 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 (Long, wo Sub)](): Higher is better; unit: PERCENT. - [Accuracy (Long, w Sub)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Long, wo Sub) | Result date | Report paper | | --- | --- | --- | --- | | Gemini-1.5-Pro | 67.4 | Not recorded | 698181f7a8b01e76cc8e29a2 | | Gemini 1.5 Pro | 67.4 | Not recorded | 694238fc4561469362850518 | | GPT4-o | 65.3 | Not recorded | 698181f7a8b01e76cc8e29a2 | | GPT-4o | 65.3 | Not recorded | 694238fc4561469362850518 | | LLaVA-OneVision (72B) | 60 | Not recorded | 698181f7a8b01e76cc8e29a2 | | InternVL3\* | 57.7 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (128f) Stage 2 | 57.7 | Not recorded | 698181f7a8b01e76cc8e29a2 | | VideoLLaMA2 (72B) | 57.6 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (128f) Stage 1 | 57 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (256f) Stage 2-iter2 | 56.1 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (256f) Stage 1 | 55.8 | Not recorded | 698181f7a8b01e76cc8e29a2 | | VideoChat-Flash | 55.4 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (128f) Stage 2 | 55.3 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (256f) Stage 2-iter1 | 54.9 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (512f) Stage 1 | 54.9 | Not recorded | 698181f7a8b01e76cc8e29a2 | | NVILA | 54.8 | Not recorded | 698181f7a8b01e76cc8e29a2 | | Zoom-Zero | 54.8 | Not recorded | 694238fc4561469362850518 | | LongVPO (256f) Stage 1 | 54.3 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVPO (128f) Stage 1 | 53.8 | Not recorded | 698181f7a8b01e76cc8e29a2 | | GPT4-V | 53.5 | Not recorded | 698181f7a8b01e76cc8e29a2 | | VideoChat-R1 | 53.4 | Not recorded | 694238fc4561469362850518 | | TVG-R1 | 52.7 | Not recorded | 694238fc4561469362850518 | | InternVideo2.5\* | 52.6 | Not recorded | 698181f7a8b01e76cc8e29a2 | | MiniCPM-v2.6 | 51.8 | Not recorded | 698181f7a8b01e76cc8e29a2 | | InternVL2.5\* | 51.1 | Not recorded | 698181f7a8b01e76cc8e29a2 | | Qwen2.5-VL | 51.1 | Not recorded | 694238fc4561469362850518 | | mPLUG-Owl3 | 50.1 | Not recorded | 698181f7a8b01e76cc8e29a2 | | Video-XL | 49.2 | Not recorded | 698181f7a8b01e76cc8e29a2 | | Video-CCAM | 46.7 | Not recorded | 698181f7a8b01e76cc8e29a2 | | Kangaroo | 46.7 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVA | 46.2 | Not recorded | 698181f7a8b01e76cc8e29a2 | | LongVA | 46.2 | Not recorded | 694238fc4561469362850518 | ### Gemini-1.5-Pro Accuracy (Long, wo Sub): 67.4. Result ID: 699757825698665ec4a9922f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"Undisclosed","Frames":null}. This record does not assert independent reproduction. ### Gemini 1.5 Pro Accuracy (Long, wo Sub): 67.4. Result ID: 69983eef5698665ec49b35e9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Proprietary"}. This record does not assert independent reproduction. ### GPT4-o Accuracy (Long, wo Sub): 65.3. Result ID: 699757825698665ec4a9922e. [Reporting paper]() Recorded evaluation configuration: {"Size":"Undisclosed","Frames":null}. This record does not assert independent reproduction. ### GPT-4o Accuracy (Long, wo Sub): 65.3. Result ID: 69983eef5698665ec49b35ea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Proprietary"}. This record does not assert independent reproduction. ### LLaVA-OneVision (72B) Accuracy (Long, wo Sub): 60. Result ID: 699757825698665ec4a99230. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"72B","Frames":"32"}. This record does not assert independent reproduction. ### InternVL3\* Accuracy (Long, wo Sub): 57.7. Result ID: 699757825698665ec4a9923a. [Reporting paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512"}. This record does not assert independent reproduction. ### LongVPO (128f) Stage 2 Accuracy (Long, wo Sub): 57.7. Result ID: 699757825698665ec4a9923c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"768","Backbone":"InternVL3"}. This record does not assert independent reproduction. ### VideoLLaMA2 (72B) Accuracy (Long, wo Sub): 57.6. Result ID: 699757825698665ec4a99234. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"72B","Frames":"16"}. This record does not assert independent reproduction. ### LongVPO (128f) Stage 1 Accuracy (Long, wo Sub): 57. Result ID: 699757825698665ec4a9923b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVL3"}. This record does not assert independent reproduction. ### LongVPO (256f) Stage 2-iter2 Accuracy (Long, wo Sub): 56.1. Result ID: 699757825698665ec4a99244. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVideo2.5"}. This record does not assert independent reproduction. ### LongVPO (256f) Stage 1 Accuracy (Long, wo Sub): 55.8. Result ID: 699757825698665ec4a99240. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVL2.5"}. This record does not assert independent reproduction. ### VideoChat-Flash Accuracy (Long, wo Sub): 55.4. Result ID: 699757825698665ec4a99239. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B","Frames":"512"}. This record does not assert independent reproduction. ### LongVPO (128f) Stage 2 Accuracy (Long, wo Sub): 55.3. Result ID: 699757825698665ec4a9923f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVL2.5"}. This record does not assert independent reproduction. ### LongVPO (256f) Stage 2-iter1 Accuracy (Long, wo Sub): 54.9. Result ID: 699757825698665ec4a99243. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVideo2.5"}. This record does not assert independent reproduction. ### LongVPO (512f) Stage 1 Accuracy (Long, wo Sub): 54.9. Result ID: 699757825698665ec4a99245. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVideo2.5"}. This record does not assert independent reproduction. ### NVILA Accuracy (Long, wo Sub): 54.8. Result ID: 699757825698665ec4a99233. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B","Frames":"256"}. This record does not assert independent reproduction. ### Zoom-Zero Accuracy (Long, wo Sub): 54.8. Result ID: 69983eef5698665ec49b35ef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source","Model Scale":"7B-8B"}. This record does not assert independent reproduction. ### LongVPO (256f) Stage 1 Accuracy (Long, wo Sub): 54.3. Result ID: 699757825698665ec4a99242. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVideo2.5"}. This record does not assert independent reproduction. ### LongVPO (128f) Stage 1 Accuracy (Long, wo Sub): 53.8. Result ID: 699757825698665ec4a9923e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"512","Backbone":"InternVL2.5"}. This record does not assert independent reproduction. ### GPT4-V Accuracy (Long, wo Sub): 53.5. Result ID: 699757825698665ec4a9922d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"Undisclosed","Frames":null}. This record does not assert independent reproduction. ### VideoChat-R1 Accuracy (Long, wo Sub): 53.4. Result ID: 69983eef5698665ec49b35ee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source","Model Scale":"7B-8B"}. This record does not assert independent reproduction. ### TVG-R1 Accuracy (Long, wo Sub): 52.7. Result ID: 69983eef5698665ec49b35ed. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source","Model Scale":"7B-8B"}. This record does not assert independent reproduction. ### InternVideo2.5\* Accuracy (Long, wo Sub): 52.6. Result ID: 699757825698665ec4a99241. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"8B","Frames":"512"}. This record does not assert independent reproduction. ### MiniCPM-v2.6 Accuracy (Long, wo Sub): 51.8. Result ID: 699757825698665ec4a99231. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"8B","Frames":"64"}. This record does not assert independent reproduction. ### InternVL2.5\* Accuracy (Long, wo Sub): 51.1. Result ID: 699757825698665ec4a9923d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"8B","Frames":"512"}. This record does not assert independent reproduction. ### Qwen2.5-VL Accuracy (Long, wo Sub): 51.1. Result ID: 69983eef5698665ec49b35ec. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Access":"Open-Source","Model Scale":"7B-8B"}. This record does not assert independent reproduction. ### mPLUG-Owl3 Accuracy (Long, wo Sub): 50.1. Result ID: 699757825698665ec4a99232. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"7B","Frames":"16"}. This record does not assert independent reproduction. ### Video-XL Accuracy (Long, wo Sub): 49.2. Result ID: 699757825698665ec4a99235. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B","Frames":"2,048"}. This record does not assert independent reproduction. ### Video-CCAM Accuracy (Long, wo Sub): 46.7. Result ID: 699757825698665ec4a99236. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"9B","Frames":"96"}. This record does not assert independent reproduction. ### Kangaroo Accuracy (Long, wo Sub): 46.7. Result ID: 699757825698665ec4a99237. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"8B","Frames":"64"}. This record does not assert independent reproduction. ### LongVA Accuracy (Long, wo Sub): 46.2. Result ID: 699757825698665ec4a99238. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"7B","Frames":"128"}. This record does not assert independent reproduction. ### LongVA Accuracy (Long, wo Sub): 46.2. Result ID: 69983eef5698665ec49b35eb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Access":"Open-Source","Model Scale":"7B-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.