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

ground truth

Water Drop

model outputs

Qwen3-VL-8B-Instruct (no output)
A T
GPT-5.4
A 0.83T 0.29
LLaMA 4 Scout
A 0.64T 0.26
1<div class="drop-container">
2  <div class="drop">
3    <h1>Protecting Our Clean Water</h1>
4    <p class="has-text-align-center has-new-text-color has-text-color">Addressing phosphorus, the greatest threat to clean water in Lake Champlain</p>
5</div>
6</div>