# Jailbreak Attack on GuidedBench

Source: [Jailbreak Attack on GuidedBench](<https://www.sota2.com/research/sota/jailbreak-attack-on-guidedbench>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:24:02.228Z

Evaluation of model robustness against jailbreak attacks using the GuidedBench dataset, measured by Attack Success Rate (ASR) using specific evaluators.

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 | Jailbreak Attack |
| Dataset (including any stated split) | GuidedBench |
| Metric | ASR (HarmBench) |
| 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

- [ASR (HarmBench)](<https://www.sota2.com/research/sota/jailbreak-attack-on-guidedbench.md?metric=ASR%20(HarmBench)>): Higher is better; unit: PERCENT.

- [ASR (GuidedEvaluator-GPT-4o-mini)](<https://www.sota2.com/research/sota/jailbreak-attack-on-guidedbench.md?metric=ASR%20(GuidedEvaluator-GPT-4o-mini)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | ASR (HarmBench) | Result date | Report paper |
| --- | --- | --- | --- |
| Original Weight | 29.5 | Not recorded | 68fa7a0d1e209131cea80d29 |
| Original Weight | 6 | Not recorded | 68fa7a0d1e209131cea80d29 |
| CB\* | 4.5 | Not recorded | 68fa7a0d1e209131cea80d29 |
| REPBEND | 3.5 | Not recorded | 68fa7a0d1e209131cea80d29 |
| REPBEND | 3 | Not recorded | 68fa7a0d1e209131cea80d29 |
| CB\* | 2.5 | Not recorded | 68fa7a0d1e209131cea80d29 |

### Original Weight

ASR (HarmBench): 29.5. Result ID: 698b6b6a7a967eddb76e6a75.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

Recorded evaluation configuration: {"Model":"Mistral 7B Instruct v0.2"}. This record does not assert independent reproduction.

### Original Weight

ASR (HarmBench): 6. Result ID: 698b6b6a7a967eddb76e6a78.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

Recorded evaluation configuration: {"Model":"Llama3 8B Instruct"}. This record does not assert independent reproduction.

### CB\*

ASR (HarmBench): 4.5. Result ID: 698b6b6a7a967eddb76e6a76.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa740d1e209131cea5f76f.md>)

[Implementation](<https://www.sota2.com/github.com/blackswan-ai/circuit-breakers>)

Recorded evaluation configuration: {"Model":"Mistral 7B Instruct v0.2"}. This record does not assert independent reproduction.

### REPBEND

ASR (HarmBench): 3.5. Result ID: 698b6b6a7a967eddb76e6a77.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

[Implementation](<https://www.sota2.com/github.com/AIM-Intelligence/RepBend>)

Recorded evaluation configuration: {"Model":"Mistral 7B Instruct v0.2"}. This record does not assert independent reproduction.

### REPBEND

ASR (HarmBench): 3. Result ID: 698b6b6a7a967eddb76e6a7a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

[Implementation](<https://www.sota2.com/github.com/AIM-Intelligence/RepBend>)

Recorded evaluation configuration: {"Model":"Llama3 8B Instruct"}. This record does not assert independent reproduction.

### CB\*

ASR (HarmBench): 2.5. Result ID: 698b6b6a7a967eddb76e6a79.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a0d1e209131cea80d29.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa740d1e209131cea5f76f.md>)

[Implementation](<https://www.sota2.com/github.com/blackswan-ai/circuit-breakers>)

Recorded evaluation configuration: {"Model":"Llama3 8B Instruct"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.