# Web Navigation and Automation on WorkArena Held-out Goals (test) Source: [Web Navigation and Automation on WorkArena Held-out Goals (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:05:27.451Z Evaluation of agent performance on the WorkArena Held-out Goals test set, focusing on web navigation and automation 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 | Web Navigation and Automation | | Dataset (including any stated split) | WorkArena Held-out Goals (test) | | Metric | Success Rate | | 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](): Higher is better; unit: PERCENT. ## Recorded results | Method | Success Rate | Result date | Report paper | | --- | --- | --- | --- | | o1-Mini | 53.8 | Not recorded | 68fa79d41e209131cea7f95b | | Claude-3.5-Sonnet | 52.5 | Not recorded | 68fa79d41e209131cea7f95b | | GPT-4o | 42.1 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.1-405B-Instruct | 39.2 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.3-70B Instruct (Teacher) | 36 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.1-8B SFT+RL (Ours) | 35.4 | Not recorded | 68fa79d41e209131cea7f95b | | Qwen-2.5-72B Instruct (Teacher) | 33.3 | Not recorded | 68fa79d41e209131cea7f95b | | Qwen-2.5-7B SFT+RL (Ours) | 32.5 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.1-8B SFT (Ours) | 28.4 | Not recorded | 68fa79d41e209131cea7f95b | | GPT-4o-Mini | 27.1 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.1-70B-Instruct | 25 | Not recorded | 68fa79d41e209131cea7f95b | | Qwen-2.5-7B SFT (Ours) | 25 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.1-8B Instruct (Student) | 8.3 | Not recorded | 68fa79d41e209131cea7f95b | | Llama-3.1-8B RL (Ours) | 8 | Not recorded | 68fa79d41e209131cea7f95b | | Qwen-2.5-7B Instruct (Student) | 5.2 | Not recorded | 68fa79d41e209131cea7f95b | | Qwen-2.5-7B RL (Ours) | 4.5 | Not recorded | 68fa79d41e209131cea7f95b | ### o1-Mini Success Rate: 53.8. Result ID: 699a631c5698665ec4eb31cc. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"o1-Mini"}. This record does not assert independent reproduction. ### Claude-3.5-Sonnet Success Rate: 52.5. Result ID: 699a631c5698665ec4eb31c8. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Claude-3.5-Sonnet"}. This record does not assert independent reproduction. ### GPT-4o Success Rate: 42.1. Result ID: 699a631c5698665ec4eb31c9. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"GPT-4o"}. This record does not assert independent reproduction. ### Llama-3.1-405B-Instruct Success Rate: 39.2. Result ID: 699a631c5698665ec4eb31cd. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.1-405B"}. This record does not assert independent reproduction. ### Llama-3.3-70B Instruct (Teacher) Success Rate: 36. Result ID: 699a631c5698665ec4eb31ce. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.3-70B","Model Role":"Teacher"}. This record does not assert independent reproduction. ### Llama-3.1-8B SFT+RL (Ours) Success Rate: 35.4. Result ID: 699a631c5698665ec4eb31d2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.1-8B","Training Protocol":"SFT+RL"}. This record does not assert independent reproduction. ### Qwen-2.5-72B Instruct (Teacher) Success Rate: 33.3. Result ID: 699a631c5698665ec4eb31d3. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Qwen-2.5-72B","Model Role":"Teacher"}. This record does not assert independent reproduction. ### Qwen-2.5-7B SFT+RL (Ours) Success Rate: 32.5. Result ID: 699a631c5698665ec4eb31d7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Qwen-2.5-7B","Training Protocol":"SFT+RL"}. This record does not assert independent reproduction. ### Llama-3.1-8B SFT (Ours) Success Rate: 28.4. Result ID: 699a631c5698665ec4eb31d1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.1-8B","Training Protocol":"SFT"}. This record does not assert independent reproduction. ### GPT-4o-Mini Success Rate: 27.1. Result ID: 699a631c5698665ec4eb31ca. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"GPT-4o-Mini"}. This record does not assert independent reproduction. ### Llama-3.1-70B-Instruct Success Rate: 25. Result ID: 699a631c5698665ec4eb31cb. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.1-70B"}. This record does not assert independent reproduction. ### Qwen-2.5-7B SFT (Ours) Success Rate: 25. Result ID: 699a631c5698665ec4eb31d6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Qwen-2.5-7B","Training Protocol":"SFT"}. This record does not assert independent reproduction. ### Llama-3.1-8B Instruct (Student) Success Rate: 8.3. Result ID: 699a631c5698665ec4eb31cf. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.1-8B","Model Role":"Student"}. This record does not assert independent reproduction. ### Llama-3.1-8B RL (Ours) Success Rate: 8. Result ID: 699a631c5698665ec4eb31d0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Llama-3.1-8B","Training Protocol":"RL"}. This record does not assert independent reproduction. ### Qwen-2.5-7B Instruct (Student) Success Rate: 5.2. Result ID: 699a631c5698665ec4eb31d4. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Qwen-2.5-7B","Model Role":"Student"}. This record does not assert independent reproduction. ### Qwen-2.5-7B RL (Ours) Success Rate: 4.5. Result ID: 699a631c5698665ec4eb31d5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Qwen-2.5-7B","Training Protocol":"RL"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.