LLM Arena

API Configuration

Endpoint URL OpenAI-compatible base URL (without /v1/chat/completions)

API Key Remember key in browser

Model

Temperature (0.7)

Seed (empty = random per run)

Request Timeout (s)

System Prompts

Blind mode

CORS Setup

This tool runs entirely in your browser and makes direct fetch() requests to LLM APIs. Your endpoint must serve appropriate CORS headers.

CORS configuration per server type
Server	CORS Behavior
Ollama (localhost:11434)	Enabled by default for localhost. Set `OLLAMA_ORIGINS=*` if needed.
LM Studio	Enable "Local Server" mode; CORS headers included automatically.
llama.cpp server	Start with `--cors` flag.
OpenAI / compatible SaaS	CORS enabled by default.
Custom proxy	Configure reverse proxy to add `Access-Control-Allow-Origin` headers.