# Prompt classification on XSTest

Source: [Prompt classification on XSTest](<https://www.sota2.com/research/sota/prompt-classification-on-xstest>)

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

Data updated: 2026-02-26T05:43:32.032Z

Evaluation of model performance on the XSTest dataset for general prompt classification tasks, measured by F1 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 | Prompt classification |
| Dataset (including any stated split) | XSTest |
| Metric | F1 Score |
| 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

- [F1 Score](<https://www.sota2.com/research/sota/prompt-classification-on-xstest.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | F1 Score | Result date | Report paper |
| --- | --- | --- | --- |
| WildGuard | 94.8 | Not recorded | 69730068b2be0ed85a78c8a4 |
| YuFeng-XGuard | 94.4 | Not recorded | 69730068b2be0ed85a78c8a4 |
| PolyGuard | 92.2 | Not recorded | 69730068b2be0ed85a78c8a4 |
| YuFeng-XGuard | 91.6 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Qwen3Guard-Gen | 90.8 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Qwen3Guard-Gen | 89.9 | Not recorded | 69730068b2be0ed85a78c8a4 |
| GPT-OSS-SafeGuard | 89.9 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Qwen3Guard-Gen | 89.3 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Llama3Guard | 88.4 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Qwen3Guard-Gen | 87.7 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Qwen3Guard-Gen | 85.7 | Not recorded | 69730068b2be0ed85a78c8a4 |
| NemotronReasoning | 84.8 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Qwen3Guard-Gen | 84.6 | Not recorded | 69730068b2be0ed85a78c8a4 |
| NemotronGuardV2 | 84.1 | Not recorded | 69730068b2be0ed85a78c8a4 |
| Llama4Guard | 83.3 | Not recorded | 69730068b2be0ed85a78c8a4 |
| ShieldGemma | 82.8 | Not recorded | 69730068b2be0ed85a78c8a4 |

### WildGuard

F1 Score: 94.8. Result ID: 699675a75698665ec4bd1e7a.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76291e209131cea6b9a6.md>)

Recorded evaluation configuration: {"Model Size":"7B"}. This record does not assert independent reproduction.

### YuFeng-XGuard

F1 Score: 94.4. Result ID: 699675a75698665ec4bd1e81.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

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

### PolyGuard

F1 Score: 92.2. Result ID: 699675a75698665ec4bd1e7d.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a751e209131cea835e9.md>)

Recorded evaluation configuration: {"Model Size":"7B"}. This record does not assert independent reproduction.

### YuFeng-XGuard

F1 Score: 91.6. Result ID: 699675a75698665ec4bd1e77.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

Recorded evaluation configuration: {"Model Size":"0.6B"}. This record does not assert independent reproduction.

### Qwen3Guard-Gen

F1 Score: 90.8. Result ID: 699675a75698665ec4bd1e80.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c351e209131cea8d09c.md>)

Recorded evaluation configuration: {"Model Size":"8B","Evaluation protocol":"strict"}. This record does not assert independent reproduction.

### Qwen3Guard-Gen

F1 Score: 89.9. Result ID: 699675a75698665ec4bd1e76.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c351e209131cea8d09c.md>)

Recorded evaluation configuration: {"Model Size":"4B","Evaluation protocol":"strict"}. This record does not assert independent reproduction.

### GPT-OSS-SafeGuard

F1 Score: 89.9. Result ID: 699675a75698665ec4bd1e7e.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ab01e209131cea84a0c.md>)

Recorded evaluation configuration: {"Model Size":"20B"}. This record does not assert independent reproduction.

### Qwen3Guard-Gen

F1 Score: 89.3. Result ID: 699675a75698665ec4bd1e7f.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c351e209131cea8d09c.md>)

Recorded evaluation configuration: {"Model Size":"8B","Evaluation protocol":"loose"}. This record does not assert independent reproduction.

### Llama3Guard

F1 Score: 88.4. Result ID: 699675a75698665ec4bd1e78.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

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

### Qwen3Guard-Gen

F1 Score: 87.7. Result ID: 699675a75698665ec4bd1e75.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c351e209131cea8d09c.md>)

Recorded evaluation configuration: {"Model Size":"4B","Evaluation protocol":"loose"}. This record does not assert independent reproduction.

### Qwen3Guard-Gen

F1 Score: 85.7. Result ID: 699675a75698665ec4bd1e73.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c351e209131cea8d09c.md>)

Recorded evaluation configuration: {"Model Size":"0.6B","Evaluation protocol":"strict"}. This record does not assert independent reproduction.

### NemotronReasoning

F1 Score: 84.8. Result ID: 699675a75698665ec4bd1e74.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78d41e209131cea7aa41.md>)

Recorded evaluation configuration: {"Model Size":"4B"}. This record does not assert independent reproduction.

### Qwen3Guard-Gen

F1 Score: 84.6. Result ID: 699675a75698665ec4bd1e72.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c351e209131cea8d09c.md>)

Recorded evaluation configuration: {"Model Size":"0.6B","Evaluation protocol":"loose"}. This record does not assert independent reproduction.

### NemotronGuardV2

F1 Score: 84.1. Result ID: 699675a75698665ec4bd1e7c.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa769b1e209131cea6e172.md>)

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

### Llama4Guard

F1 Score: 83.3. Result ID: 699675a75698665ec4bd1e79.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Model Size":"12B"}. This record does not assert independent reproduction.

### ShieldGemma

F1 Score: 82.8. Result ID: 699675a75698665ec4bd1e7b.

[Reporting paper](<https://www.sota2.com/research/paper/69730068b2be0ed85a78c8a4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74661e209131cea617ba.md>)

Recorded evaluation configuration: {"Model Size":"9B"}. This record does not assert independent reproduction.

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