# Video Question Answering on StreamGaze Source: [Video Question Answering on StreamGaze]() Collection index: [llms.txt]() Data updated: 2026-03-30T06:49:22.671Z Evaluation of model performance on the StreamGaze dataset for Video Question Answering. 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 Question Answering | | Dataset (including any stated split) | StreamGaze | | Metric | NFI Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 15 of 15 | | 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 - [NFI Score](): Higher is better; unit: PERCENT. - [Overall Accuracy](): Higher is better; unit: PERCENT. - [OTP Score](): Higher is better; unit: PERCENT. - [SR Score](): Higher is better; unit: PERCENT. - [GSM Score](): Higher is better; unit: PERCENT. - [OI (E) Score](): Higher is better; unit: PERCENT. - [OI (H) Score](): Higher is better; unit: PERCENT. - [OAR Score](): Higher is better; unit: PERCENT. - [FAP Score](): Higher is better; unit: PERCENT. - [GTA Score](): Higher is better; unit: PERCENT. - [OAA Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | NFI Score | Result date | Report paper | | --- | --- | --- | --- | | Human | 70 | Not recorded | 69ca03980636e5bf4afc93c5 | | GazeQwen | 65.7 | Not recorded | 69ca03980636e5bf4afc93c5 | | GPT-4o | 60.1 | Not recorded | 69ca03980636e5bf4afc93c5 | | Qwen2.5-VL | 51.8 | Not recorded | 69ca03980636e5bf4afc93c5 | | Claude Sonnet4 | 50 | Not recorded | 69ca03980636e5bf4afc93c5 | | InternVL3.5 | 49 | Not recorded | 69ca03980636e5bf4afc93c5 | | VITA 1.5 | 47.4 | Not recorded | 69ca03980636e5bf4afc93c5 | | ViSpeak | 46.3 | Not recorded | 69ca03980636e5bf4afc93c5 | | MiniCPM-V | 43 | Not recorded | 69ca03980636e5bf4afc93c5 | | Claude Opus4 | 37.2 | Not recorded | 69ca03980636e5bf4afc93c5 | | Dispider | 36.6 | Not recorded | 69ca03980636e5bf4afc93c5 | | Kangaroo | 36.3 | Not recorded | 69ca03980636e5bf4afc93c5 | | AssistGaze | 29.4 | Not recorded | 69ca03980636e5bf4afc93c5 | | Flash-VStream | 24.9 | Not recorded | 69ca03980636e5bf4afc93c5 | | VideoLLM-online | 0 | Not recorded | 69ca03980636e5bf4afc93c5 | ### Human NFI Score: 70. Result ID: 69ca1c8947f34ce924b22150. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GazeQwen NFI Score: 65.7. Result ID: 69ca1c8947f34ce924b2215e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Frames":"Adapt."}. This record does not assert independent reproduction. ### GPT-4o NFI Score: 60.1. Result ID: 69ca1c8947f34ce924b22151. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### Qwen2.5-VL NFI Score: 51.8. Result ID: 69ca1c8947f34ce924b22155. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Frames":"Adapt."}. This record does not assert independent reproduction. ### Claude Sonnet4 NFI Score: 50. Result ID: 69ca1c8947f34ce924b22152. [Reporting paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### InternVL3.5 NFI Score: 49. Result ID: 69ca1c8947f34ce924b22156. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"8B","Frames":"Adapt."}. This record does not assert independent reproduction. ### VITA 1.5 NFI Score: 47.4. Result ID: 69ca1c8947f34ce924b22157. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Frames":"16"}. This record does not assert independent reproduction. ### ViSpeak NFI Score: 46.3. Result ID: 69ca1c8947f34ce924b2215a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Params":"7B","Frames":"1 fps"}. This record does not assert independent reproduction. ### MiniCPM-V NFI Score: 43. Result ID: 69ca1c8947f34ce924b22158. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"8B","Frames":"32"}. This record does not assert independent reproduction. ### Claude Opus4 NFI Score: 37.2. Result ID: 69ca1c8947f34ce924b22153. [Reporting paper]() Recorded evaluation configuration: {"Frames":"16"}. This record does not assert independent reproduction. ### Dispider NFI Score: 36.6. Result ID: 69ca1c8947f34ce924b2215b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B","Frames":"1 fps"}. This record does not assert independent reproduction. ### Kangaroo NFI Score: 36.3. Result ID: 69ca1c8947f34ce924b22159. [Reporting paper]() Recorded evaluation configuration: {"Params":"7B","Frames":"64"}. This record does not assert independent reproduction. ### AssistGaze NFI Score: 29.4. Result ID: 69ca1c8947f34ce924b22154. [Reporting paper]() Recorded evaluation configuration: {"Params":"26M","Frames":"32"}. This record does not assert independent reproduction. ### Flash-VStream NFI Score: 24.9. Result ID: 69ca1c8947f34ce924b2215c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"7B","Frames":"1 fps"}. This record does not assert independent reproduction. ### VideoLLM-online NFI Score: 0. Result ID: 69ca1c8947f34ce924b2215d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Params":"8B","Frames":"2 fps"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.