# Jailbreaking on AdvBench (ASR, Toxicity, WASR, W-Toxicity Scores) Source: [Jailbreaking on AdvBench (ASR, Toxicity, WASR, W-Toxicity Scores)]() Collection index: [llms.txt]() Data updated: 2026-05-19T09:31:56.732Z Evaluation of model susceptibility to jailbreaking prompts using the AdvBench dataset, reporting Attack Success Rate (ASR), Toxicity Score, Weighted Attack Success Rate (WASR), and Weighted Toxicity Score. 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 | Jailbreaking | | Dataset (including any stated split) | AdvBench | | Metric | ASR | | 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 - [ASR](): Higher is better; unit: PERCENT. - [Toxicity Score](): Lower is better; unit: PERCENT. - [WASR](): Higher is better; unit: PERCENT. - [W-Toxicity Score](): Lower is better; unit: PERCENT. ## Recorded results | Method | ASR | Result date | Report paper | | --- | --- | --- | --- | | wide-net-casting jailbreak method | 61.2 | Not recorded | 6a0beec1dbc836b87d82c63a | | wide-net-casting jailbreak method | 55.2 | Not recorded | 6a0beec1dbc836b87d82c63a | | Baseline (MLAI + PixArt-α) | 52.1 | Not recorded | 6a0beec1dbc836b87d82c63a | | wide-net-casting jailbreak method | 51.2 | Not recorded | 6a0beec1dbc836b87d82c63a | | Baseline (MLAI + PixArt-α) | 47.7 | Not recorded | 6a0beec1dbc836b87d82c63a | | wide-net-casting jailbreak method | 46.1 | Not recorded | 6a0beec1dbc836b87d82c63a | | Baseline (MLAI + PixArt-α) | 43.6 | Not recorded | 6a0beec1dbc836b87d82c63a | | Baseline (MLAI + PixArt-α) | 38.9 | Not recorded | 6a0beec1dbc836b87d82c63a | ### wide-net-casting jailbreak method ASR: 61.2. Result ID: 6a0c2d63fbc373592c0295a5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"GLM-4V-Plus","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### wide-net-casting jailbreak method ASR: 55.2. Result ID: 6a0c2d63fbc373592c0295a7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"GPT-4o","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### Baseline (MLAI + PixArt-α) ASR: 52.1. Result ID: 6a0c2d63fbc373592c0295a0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"GLM-4V-Plus","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### wide-net-casting jailbreak method ASR: 51.2. Result ID: 6a0c2d63fbc373592c0295a4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"Qwen-VL-Max","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### Baseline (MLAI + PixArt-α) ASR: 47.7. Result ID: 6a0c2d63fbc373592c0295a2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"GPT-4o","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### wide-net-casting jailbreak method ASR: 46.1. Result ID: 6a0c2d63fbc373592c0295a6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"Gemini-1.5-Pro","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### Baseline (MLAI + PixArt-α) ASR: 43.6. Result ID: 6a0c2d63fbc373592c02959f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"Qwen-VL-Max","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. ### Baseline (MLAI + PixArt-α) ASR: 38.9. Result ID: 6a0c2d63fbc373592c0295a1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target Model":"Gemini-1.5-Pro","Evaluation Scenario":"wide-net-casting"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.