# Mobile UI Task Automation on AndroidWorld (full) Source: [Mobile UI Task Automation on AndroidWorld (full)]() Collection index: [llms.txt]() Data updated: 2026-05-12T15:32:53.642Z Evaluation of model performance on the full AndroidWorld dataset for Mobile UI Task Automation, reporting success rates across Easy, Medium, and Hard scenarios, alongside efficiency metrics like Tokens/Task, Mean Steps, and Run Time. 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 | Mobile UI Task Automation | | Dataset (including any stated split) | AndroidWorld (full) | | Metric | Success Rate (Easy) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [Success Rate (Easy)](): Higher is better; unit: PERCENT. - [Success Rate (Medium)](): Higher is better; unit: PERCENT. - [Success Rate (Hard)](): Higher is better; unit: PERCENT. - [Overall Success Rate](): Higher is better; unit: PERCENT. - [Tokens/Task](): Lower is better; unit: SCALAR. - [Mean Steps](): Lower is better; unit: SCALAR. - [Run Time](): Lower is better; unit: TIME. ## Recorded results | Method | Success Rate (Easy) | Result date | Report paper | | --- | --- | --- | --- | | Gemini-3-Flash | 82.79 | Not recorded | 6a02b478dbc836b87d824bcf | | Gemini-3-Flash | 81.15 | Not recorded | 6a02b478dbc836b87d824bcf | | GPT-5.4 | 76.23 | Not recorded | 6a02b478dbc836b87d824bcf | | GPT-5.4 | 72.68 | Not recorded | 6a02b478dbc836b87d824bcf | | Gemini-3-Flash | 72.13 | Not recorded | 6a02b478dbc836b87d824bcf | | GPT-5.4 | 72.13 | Not recorded | 6a02b478dbc836b87d824bcf | | GPT-5.4 | 69.67 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-32B | 64.75 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-32B | 63.11 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-32B | 63.11 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-32B | 62.57 | Not recorded | 6a02b478dbc836b87d824bcf | | Gemini-3-Flash | 62.29 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-8B | 59.01 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-8B | 54.92 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-8B | 54.64 | Not recorded | 6a02b478dbc836b87d824bcf | | Qwen3-VL-8B | 50.82 | Not recorded | 6a02b478dbc836b87d824bcf | ### Gemini-3-Flash Success Rate (Easy): 82.79. Result ID: 6a0347ce47f34ce924e872ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Full Text"}. This record does not assert independent reproduction. ### Gemini-3-Flash Success Rate (Easy): 81.15. Result ID: 6a0347ce47f34ce924e872b9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Delta Text"}. This record does not assert independent reproduction. ### GPT-5.4 Success Rate (Easy): 76.23. Result ID: 6a0347ce47f34ce924e872bd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Delta Text"}. This record does not assert independent reproduction. ### GPT-5.4 Success Rate (Easy): 72.68. Result ID: 6a0347ce47f34ce924e872bc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"None (Baseline)"}. This record does not assert independent reproduction. ### Gemini-3-Flash Success Rate (Easy): 72.13. Result ID: 6a0347ce47f34ce924e872bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Code2Image"}. This record does not assert independent reproduction. ### GPT-5.4 Success Rate (Easy): 72.13. Result ID: 6a0347ce47f34ce924e872be. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Full Text"}. This record does not assert independent reproduction. ### GPT-5.4 Success Rate (Easy): 69.67. Result ID: 6a0347ce47f34ce924e872bf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Code2Image"}. This record does not assert independent reproduction. ### Qwen3-VL-32B Success Rate (Easy): 64.75. Result ID: 6a0347ce47f34ce924e872b5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Delta Text"}. This record does not assert independent reproduction. ### Qwen3-VL-32B Success Rate (Easy): 63.11. Result ID: 6a0347ce47f34ce924e872b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Full Text"}. This record does not assert independent reproduction. ### Qwen3-VL-32B Success Rate (Easy): 63.11. Result ID: 6a0347ce47f34ce924e872b7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Code2Image"}. This record does not assert independent reproduction. ### Qwen3-VL-32B Success Rate (Easy): 62.57. Result ID: 6a0347ce47f34ce924e872b4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"None (Baseline)"}. This record does not assert independent reproduction. ### Gemini-3-Flash Success Rate (Easy): 62.29. Result ID: 6a0347ce47f34ce924e872b8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"None (Baseline)"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Success Rate (Easy): 59.01. Result ID: 6a0347ce47f34ce924e872b1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Delta Text"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Success Rate (Easy): 54.92. Result ID: 6a0347ce47f34ce924e872b3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Code2Image"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Success Rate (Easy): 54.64. Result ID: 6a0347ce47f34ce924e872b0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"None (Baseline)"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Success Rate (Easy): 50.82. Result ID: 6a0347ce47f34ce924e872b2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Agent Framework":"M3A","World-model Feedback":"Full Text"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.