# Visual Question Answering on RoboFine-Bench (test) Source: [Visual Question Answering on RoboFine-Bench (test)]() Collection index: [llms.txt]() Data updated: 2026-05-27T10:21:11.681Z Evaluation of model performance on the test split of the RoboFine-Bench dataset for Visual Question Answering, reporting detailed metrics including Overall, Actor/Object interactions, and Configuration success rates. 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) | RoboFine-Bench (test) | | Metric | Overall Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 6 of 6 | | 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 - [Overall Accuracy](): Higher is better; unit: PERCENT. - [AA (Active Actor) Success Rate](): Higher is better; unit: PERCENT. - [TO (Target Object) Success Rate](): Higher is better; unit: PERCENT. - [IC (Initial Configuration) Success Rate](): Higher is better; unit: PERCENT. - [AS (Action Sequence) Success Rate](): Higher is better; unit: PERCENT. - [C&A (Contact & Approach) Success Rate](): Higher is better; unit: PERCENT. - [T&O (Trajectory & Orientation) Success Rate](): Higher is better; unit: PERCENT. - [BM (Body Motion) Success Rate](): Higher is better; unit: PERCENT. - [OI (Object Interaction) Success Rate](): Higher is better; unit: PERCENT. - [FC (Final Configuration) Success Rate](): Higher is better; unit: PERCENT. - [F&R (Failure & Recovery) Success Rate](): Higher is better; unit: PERCENT. ## Recorded results | Method | Overall Accuracy | Result date | Report paper | | --- | --- | --- | --- | | RoboFine-VLM | 71 | Not recorded | 6a167ad7dbd7150506f802e4 | | Gemini-3.1-Pro | 62.1 | Not recorded | 6a167ad7dbd7150506f802e4 | | GPT-5.4 | 61 | Not recorded | 6a167ad7dbd7150506f802e4 | | Doubao-Seed-2.0-Pro | 54.9 | Not recorded | 6a167ad7dbd7150506f802e4 | | Qwen3.5-Plus | 52.6 | Not recorded | 6a167ad7dbd7150506f802e4 | | Qwen3-VL-Plus | 50.4 | Not recorded | 6a167ad7dbd7150506f802e4 | ### RoboFine-VLM Overall Accuracy: 71. Result ID: 6a16c48cfbc373592c1085ed. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Gemini-3.1-Pro Overall Accuracy: 62.1. Result ID: 6a16c48cfbc373592c1085eb. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-5.4 Overall Accuracy: 61. Result ID: 6a16c48cfbc373592c1085ec. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Doubao-Seed-2.0-Pro Overall Accuracy: 54.9. Result ID: 6a16c48cfbc373592c1085ea. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen3.5-Plus Overall Accuracy: 52.6. Result ID: 6a16c48cfbc373592c1085e9. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen3-VL-Plus Overall Accuracy: 50.4. Result ID: 6a16c48cfbc373592c1085e8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.