# Long Video Question Answering on Video-MME Source: [Long Video Question Answering on Video-MME]() Collection index: [llms.txt]() Data updated: 2026-05-25T09:08:09.287Z Evaluation of Video Question Answering performance on the Video-MME dataset, reporting scores segmented by video length (Short, Medium, Long) and subtitle availability. 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 Question Answering | | Dataset (including any stated split) | Video-MME | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 53 of 53 | | 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. - [Accuracy (Short, w/o subs)](): Higher is better; unit: PERCENT. - [Accuracy (Short, w/ subs)](): Higher is better; unit: PERCENT. - [Accuracy (Medium, w/o subs)](): Higher is better; unit: PERCENT. - [Accuracy (Medium, w/ subs)](): Higher is better; unit: PERCENT. - [Accuracy (Long, w/o subs)](): Higher is better; unit: PERCENT. - [Accuracy (Long, w/ subs)](): Higher is better; unit: PERCENT. - [Average Accuracy (w/o subs)](): Higher is better; unit: PERCENT. - [Average Accuracy (w/ subs)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | ToolMerge | 73.2 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | WFS (CVPR 2026) | 72.6 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | AKS (CVPR 2025) | 72.5 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | BOLT (CVPR 2025) | 71.7 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | ToolMerge | 71 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | Uniform | 70.6 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | ToolMerge | 70.6 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | SigLIP-Q | 69.2 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | SigLIP-Q | 68 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | WFS (CVPR 2026) | 67.5 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | BOLT (CVPR 2025) | 67.4 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | Uniform | 67.3 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | AKS (CVPR 2025) | 67.3 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | LIF (NeurIPS 2025) | 67.2 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | BOLT (CVPR 2025) | 67 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | WFS (CVPR 2026) | 66.6 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | AKS (CVPR 2025) | 64.7 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | ToolMerge | 64.6 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | SigLIP-Q | 64.2 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | Uniform | 64 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | LIF (NeurIPS 2025) | 63.6 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | WFS (CVPR 2026) | 63.3 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | LIF (NeurIPS 2025) | 62.8 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | BOLT (CVPR 2025) | 62.4 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | SigLIP-Q | 58.5 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | AKS (CVPR 2025) | 58.5 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | Uniform | 57.9 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | LIF (NeurIPS 2025) | 57 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | Blind Text (0 frames) | 44.5 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | | Blind Text (0 frames) | 38.7 | Not recorded | 6a13d7ac6f71c5ae7dc9ea96 | ### ToolMerge Accuracy: 73.2. Result ID: 6a14115ffbc373592c0b7663. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### WFS (CVPR 2026) Accuracy: 72.6. Result ID: 6a14115ffbc373592c0b765f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### AKS (CVPR 2025) Accuracy: 72.5. Result ID: 6a14115ffbc373592c0b7653. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### BOLT (CVPR 2025) Accuracy: 71.7. Result ID: 6a14115ffbc373592c0b7657. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### ToolMerge Accuracy: 71. Result ID: 6a14115ffbc373592c0b7662. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### Uniform Accuracy: 70.6. Result ID: 6a14115ffbc373592c0b764b. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### ToolMerge Accuracy: 70.6. Result ID: 6a14115ffbc373592c0b7661. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### SigLIP-Q Accuracy: 69.2. Result ID: 6a14115ffbc373592c0b764f. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### SigLIP-Q Accuracy: 68. Result ID: 6a14115ffbc373592c0b764d. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### WFS (CVPR 2026) Accuracy: 67.5. Result ID: 6a14115ffbc373592c0b765d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### BOLT (CVPR 2025) Accuracy: 67.4. Result ID: 6a14115ffbc373592c0b7655. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### Uniform Accuracy: 67.3. Result ID: 6a14115ffbc373592c0b7649. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### AKS (CVPR 2025) Accuracy: 67.3. Result ID: 6a14115ffbc373592c0b7651. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### LIF (NeurIPS 2025) Accuracy: 67.2. Result ID: 6a14115ffbc373592c0b765b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### BOLT (CVPR 2025) Accuracy: 67. Result ID: 6a14115ffbc373592c0b7656. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### WFS (CVPR 2026) Accuracy: 66.6. Result ID: 6a14115ffbc373592c0b765e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### AKS (CVPR 2025) Accuracy: 64.7. Result ID: 6a14115ffbc373592c0b7652. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### ToolMerge Accuracy: 64.6. Result ID: 6a14115ffbc373592c0b7660. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### SigLIP-Q Accuracy: 64.2. Result ID: 6a14115ffbc373592c0b764e. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### Uniform Accuracy: 64. Result ID: 6a14115ffbc373592c0b764a. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### LIF (NeurIPS 2025) Accuracy: 63.6. Result ID: 6a14115ffbc373592c0b765a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### WFS (CVPR 2026) Accuracy: 63.3. Result ID: 6a14115ffbc373592c0b765c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### LIF (NeurIPS 2025) Accuracy: 62.8. Result ID: 6a14115ffbc373592c0b7659. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"32"}. This record does not assert independent reproduction. ### BOLT (CVPR 2025) Accuracy: 62.4. Result ID: 6a14115ffbc373592c0b7654. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### SigLIP-Q Accuracy: 58.5. Result ID: 6a14115ffbc373592c0b764c. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### AKS (CVPR 2025) Accuracy: 58.5. Result ID: 6a14115ffbc373592c0b7650. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### Uniform Accuracy: 57.9. Result ID: 6a14115ffbc373592c0b7648. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### LIF (NeurIPS 2025) Accuracy: 57. Result ID: 6a14115ffbc373592c0b7658. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"8"}. This record does not assert independent reproduction. ### Blind Text (0 frames) Accuracy: 44.5. Result ID: 6a14115ffbc373592c0b7646. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"Qwen3-VL","Frame budget (K)":"0"}. This record does not assert independent reproduction. ### Blind Text (0 frames) Accuracy: 38.7. Result ID: 6a14115ffbc373592c0b7647. [Reporting paper]() Recorded evaluation configuration: {"Answerer model":"GPT-4o","Frame budget (K)":"0"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.