Fable 5 AI Playground
Test your prompts on real, free open-weight models. The planner on the home page structures your idea — the Playground shows you what a live model actually does with it.
Turn a rough idea into a structured, reusable prompt
Output comes from third-party open-weight models routed through OpenRouter — not from an official “Fable 5” model. AI output can be wrong; verify before relying on it. Don’t submit personal or confidential information.
How the Playground works
1. Pick a mode and model
Four modes cover the common jobs: refining a prompt, drafting creative text, structured analysis, and prompt-safety review. Models are curated free routes from the model directory.
2. Run on a live model
Your prompt is sent server-side to OpenRouter with automatic fallback: if one free provider is busy, the request is rerouted to the next model in the list, and the output header shows which model answered.
3. Save what works
When a refined prompt performs well, keep it. The template library has 40+ proven scaffolds you can adapt instead of starting from scratch.
Transparency: Anthropic’s Fable 5 is a paid model ($10/$50 per million tokens via the Claude API), so this free Playground runs open-weight models (GPT-OSS, Kimi, Llama, Qwen, GLM) through OpenRouter instead — use it to refine prompt structure, then run the finished prompt in your own Claude app or API account. Requests are rate-limited per visitor to keep the service free. See the privacy policy for how prompt data is handled.