Set the percentage of tests to run during the benchmark. 100% runs all tests.

<!-- CONFIG_START -->
RUN_PERCENTAGE: 100
SHARED_PROMPT: "Provide production-ready and maintainable JavaScript code. Apply code golfing practices but don't put everything in a single line. No comments. Your code will execute in the browser."
<!-- CONFIG_END -->


The following models are included in the benchmark run.

<!-- MODELS_START -->
openai/gpt-5.1-codex
openai/gpt-5.1-chat
google/gemini-2.5-pro
anthropic/claude-sonnet-4.5 TEMP:0.7
<!-- MODELS_END -->