animation2code benchmark
For best compatibility, please view this dashboard in a Chrome browser.
← back to model results

ground truth

flat loading animate

model outputs

Qwen3-VL-8B-Instruct (no output)
A T
GPT-5.4
A 0.95T 0.24
LLaMA 4 Scout
A 0.56T 0.18
1<div>
2  <div class="loading4"><i></i><i></i><i></i><i></i></div>
3</div>
4