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.

Optional flags after a model name: TEMP:<n> sets temperature, EFF:<level> sets reasoning effort.
API:anthropic calls the Anthropic API directly (ANTHROPIC_API_KEY) instead of OpenRouter. Keep the
OpenRouter name; the Anthropic model id is that name without "anthropic/" and with dots turned into
dashes (anthropic/claude-fable-5.2 -> claude-fable-5-2). The flag is not shown on the site or in output
file names, and TEMP is ignored with it.

<!-- MODELS_START -->
anthropic/claude-fable-5.2 EFF:xhigh API:anthropic
anthropic/claude-opus-5.5 EFF:xhigh API:anthropic
google/gemini-3.8-flash EFF:high
google/gemini-3.7-flash EFF:high
moonshotai/kimi-k3 EFF:medium
openai/gpt-5.6-luna EFF:high
x-ai/grok-4.5
anthropic/claude-sonnet-5 EFF:high
z-ai/glm-5.2
anthropic/claude-fable-5
anthropic/claude-opus-4.8 EFF:medium
google/gemini-3.5-flash
google/gemini-3.1-flash-lite
openai/gpt-5.5 EFF:high
moonshotai/kimi-k2.6
anthropic/claude-opus-4.7 EFF:medium
google/gemini-3.1-pro-preview
google/gemini-3-flash-preview TEMP:0.35
deepseek/deepseek-v3.2
<!-- MODELS_END -->
S
Description
LLM Benchmark
Readme
3 MiB
0 Stars 1 Watchers 0 Forks
Languages
JavaScript 94.5%
HTML 5.5%