Compare two LLM responses side by side with textual diff, semantic similarity scoring, and performance metrics.
Frontend only — No backend, no auth, no server. Runs entirely in the browser.
- React + TypeScript
- Tailwind CSS
- Vite
- Transformers.js (client-side embeddings)
- IndexedDB (local persistence)
npm install
npm run devnpm run buildAPI keys for LLM providers (OpenAI, Anthropic, Mistral) are stored in localStorage on the client side. They never transit through a server. This is a known limitation for a local-first tool.