# Website Accessibility Correction on WebAccessVL (test)

Source: [Website Accessibility Correction on WebAccessVL (test)](<https://www.sota2.com/research/sota/website-accessibility-correction-on-webaccessvl-test>)

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

Data updated: 2026-02-26T06:08:17.172Z

Evaluation of models on the WebAccessVL dataset (test split) for Website Accessibility Correction, measured by metrics including structural accuracy, tree edit distance, and CLIP similarity.

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 | Website Accessibility Correction |
| Dataset (including any stated split) | WebAccessVL (test) |
| Metric | Violations |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 18 of 18 |
| 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

- [Violations](<https://www.sota2.com/research/sota/website-accessibility-correction-on-webaccessvl-test.md?metric=Violations>): Lower is better; unit: SCALAR.

- [Caption-Image CLIP Similarity](<https://www.sota2.com/research/sota/website-accessibility-correction-on-webaccessvl-test.md?metric=Caption-Image%20CLIP%20Similarity>): Higher is better; unit: SCALAR.

- [Structural Accuracy](<https://www.sota2.com/research/sota/website-accessibility-correction-on-webaccessvl-test.md?metric=Structural%20Accuracy>): Higher is better; unit: PERCENT.

- [Tree Edit Distance](<https://www.sota2.com/research/sota/website-accessibility-correction-on-webaccessvl-test.md?metric=Tree%20Edit%20Distance>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Violations | Result date | Report paper |
| --- | --- | --- | --- |
| Llama 3.2 Vision | 0.439 | Not recorded | 698423eea8b01e76cc8ee610 |
| LLaVA 1.6 | 0.509 | Not recorded | 698423eea8b01e76cc8ee610 |
| GPT-5 | 1.681 | Not recorded | 698423eea8b01e76cc8ee610 |
| Claude3.5 Sonnet | 2.31 | Not recorded | 698423eea8b01e76cc8ee610 |
| GPT-4o Vision | 2.918 | Not recorded | 698423eea8b01e76cc8ee610 |
| GPT-4o | 2.929 | Not recorded | 698423eea8b01e76cc8ee610 |
| Qwen 2.5 VL 7B | 3.77 | Not recorded | 698423eea8b01e76cc8ee610 |
| Llama 3.2 3B | 4.382 | Not recorded | 698423eea8b01e76cc8ee610 |
| Qwen 2.5 Coder 7B | 4.466 | Not recorded | 698423eea8b01e76cc8ee610 |
| Llama 3.2 Vision 11B | 4.765 | Not recorded | 698423eea8b01e76cc8ee610 |
| LLaVA 1.5 | 4.944 | Not recorded | 698423eea8b01e76cc8ee610 |
| Qwen 2 7B | 5.173 | Not recorded | 698423eea8b01e76cc8ee610 |
| LLaVA 1.6 | 5.203 | Not recorded | 698423eea8b01e76cc8ee610 |
| Qwen 2 VL 7B | 5.214 | Not recorded | 698423eea8b01e76cc8ee610 |
| Qwen 2.5 7B | 5.303 | Not recorded | 698423eea8b01e76cc8ee610 |
| Janus Pro 7B | 5.335 | Not recorded | 698423eea8b01e76cc8ee610 |
| DeepSeek VL 2 small | 5.505 | Not recorded | 698423eea8b01e76cc8ee610 |
| Gemini 1.5 Flash | 6.213 | Not recorded | 698423eea8b01e76cc8ee610 |

### Llama 3.2 Vision

Violations: 0.439. Result ID: 69989bf85698665ec4fe500f.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":true,"Training":"Violation-conditioned"}. This record does not assert independent reproduction.

### LLaVA 1.6

Violations: 0.509. Result ID: 69989bf85698665ec4fe500e.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":true,"Training":"Violation-conditioned"}. This record does not assert independent reproduction.

### GPT-5

Violations: 1.681. Result ID: 69989bf85698665ec4fe4fff.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### Claude3.5 Sonnet

Violations: 2.31. Result ID: 69989bf85698665ec4fe5000.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### GPT-4o Vision

Violations: 2.918. Result ID: 69989bf85698665ec4fe5006.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### GPT-4o

Violations: 2.929. Result ID: 69989bf85698665ec4fe4ffe.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### Qwen 2.5 VL 7B

Violations: 3.77. Result ID: 69989bf85698665ec4fe500c.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### Llama 3.2 3B

Violations: 4.382. Result ID: 69989bf85698665ec4fe5002.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

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

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

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### Qwen 2.5 Coder 7B

Violations: 4.466. Result ID: 69989bf85698665ec4fe5005.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75b71e209131cea6981e.md>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### Llama 3.2 Vision 11B

Violations: 4.765. Result ID: 69989bf85698665ec4fe5009.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

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

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

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### LLaVA 1.5

Violations: 4.944. Result ID: 69989bf85698665ec4fe500a.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa732e1e209131cea5af59.md>)

[Implementation](<https://llava-vl.github.io/>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### Qwen 2 7B

Violations: 5.173. Result ID: 69989bf85698665ec4fe5003.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74bf1e209131cea6378d.md>)

[Implementation](<https://github.com/QwenLM/Qwen2>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### LLaVA 1.6

Violations: 5.203. Result ID: 69989bf85698665ec4fe500b.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### Qwen 2 VL 7B

Violations: 5.214. Result ID: 69989bf85698665ec4fe500d.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75151e209131cea65976.md>)

[Implementation](<https://github.com/QwenLM/Qwen2-VL>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### Qwen 2.5 7B

Violations: 5.303. Result ID: 69989bf85698665ec4fe5004.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa767e1e209131cea6d6a7.md>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

### Janus Pro 7B

Violations: 5.335. Result ID: 69989bf85698665ec4fe5008.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76cc1e209131cea6f20a.md>)

[Implementation](<https://github.com/deepseek-ai/Janus>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### DeepSeek VL 2 small

Violations: 5.505. Result ID: 69989bf85698665ec4fe5007.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa763a1e209131cea6bfb0.md>)

Recorded evaluation configuration: {"VLM":true,"Fine-tuned":false}. This record does not assert independent reproduction.

### Gemini 1.5 Flash

Violations: 6.213. Result ID: 69989bf85698665ec4fe5001.

[Reporting paper](<https://www.sota2.com/research/paper/698423eea8b01e76cc8ee610.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76361e209131cea6be11.md>)

Recorded evaluation configuration: {"VLM":false,"Fine-tuned":false}. This record does not assert independent reproduction.

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