# Visual Question Answering on IRIS 1.0 (ambiguous trials) Source: [Visual Question Answering on IRIS 1.0 (ambiguous trials)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:34:17.931Z Evaluation of model performance on the IRIS 1.0 (ambiguous trials) dataset for Visual Question Answering, reporting accuracies based on image input only versus image and gaze input. 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 | Visual Question Answering | | Dataset (including any stated split) | IRIS 1.0 (ambiguous trials) | | Metric | Accuracy (Image Only) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 (Image Only)](): Higher is better; unit: PERCENT. - [Accuracy (Image+Gaze)](): Higher is better; unit: PERCENT. - [Delta (Percentage Points)](): Higher is better; unit: SCALAR. - [p-value](): Lower is better; unit: SCALAR. ## Recorded results | Method | Accuracy (Image Only) | Result date | Report paper | | --- | --- | --- | --- | | Gemini 2.5 Pro | 59.3 | Not recorded | 699698e909c33900fcd71d94 | | Qwen 2.5 VL 32B | 56.2 | Not recorded | 699698e909c33900fcd71d94 | | Claude 4.1 Opus | 55.2 | Not recorded | 699698e909c33900fcd71d94 | | Gemini 2.5 Flash | 54.2 | Not recorded | 699698e909c33900fcd71d94 | | Claude Sonnet 4 | 54.2 | Not recorded | 699698e909c33900fcd71d94 | | GPT-5 | 53.7 | Not recorded | 699698e909c33900fcd71d94 | | Ovis 2.5 9B | 52.5 | Not recorded | 699698e909c33900fcd71d94 | | Qwen 2.5 VL 3B | 52 | Not recorded | 699698e909c33900fcd71d94 | | Qwen 2.5 VL 7B | 51.7 | Not recorded | 699698e909c33900fcd71d94 | | GPT-5 Mini | 49.7 | Not recorded | 699698e909c33900fcd71d94 | ### Gemini 2.5 Pro Accuracy (Image Only): 59.3. Result ID: 699aec225698665ec47ce1de. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Qwen 2.5 VL 32B Accuracy (Image Only): 56.2. Result ID: 699aec225698665ec47ce1e3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Claude 4.1 Opus Accuracy (Image Only): 55.2. Result ID: 699aec225698665ec47ce1e2. [Reporting paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Gemini 2.5 Flash Accuracy (Image Only): 54.2. Result ID: 699aec225698665ec47ce1dd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Claude Sonnet 4 Accuracy (Image Only): 54.2. Result ID: 699aec225698665ec47ce1e0. [Reporting paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### GPT-5 Accuracy (Image Only): 53.7. Result ID: 699aec225698665ec47ce1df. [Reporting paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Ovis 2.5 9B Accuracy (Image Only): 52.5. Result ID: 699aec225698665ec47ce1e1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Qwen 2.5 VL 3B Accuracy (Image Only): 52. Result ID: 699aec225698665ec47ce1e5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### Qwen 2.5 VL 7B Accuracy (Image Only): 51.7. Result ID: 699aec225698665ec47ce1e4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. ### GPT-5 Mini Accuracy (Image Only): 49.7. Result ID: 699aec225698665ec47ce1dc. [Reporting paper]() Recorded evaluation configuration: {"Temporal Window":"±1s around speech onset"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.