A prompt workbench for your desktop.
Save and version your prompts, run one prompt side by side on Claude, GPT, Gemini and local Ollama models, and evaluate it over a whole dataset. Bring your own API keys: they stay on your computer.
- A library with folders, tags, favorites, typo-tolerant search and real version diffs
- Streaming side-by-side runs with token counts and cost estimates
- Batch runs, ratings and version comparison (Arcana Pro)
- Sync between your computers and public share links (Arcana Pro)