Playground
Try OpenAI, Anthropic, Google and 200+ other models directly in the OrcaRouter playground. No credit card required; sign in to save sessions, compare models side-by-side, and inspect the routing decision and per-token cost for every request.
What you can do
- Send a prompt to any of 200+ frontier and open-weights models without writing any code or installing an SDK.
- Switch model in one click to compare outputs without rewriting the prompt.
- See the live token cost, latency, and chosen upstream provider for every request.
- Save sessions, share links, and export conversations as JSON, Markdown, or OpenAI-compatible request payloads.
When to use the playground
The playground is the fastest path from "I read about a new model" to "I have a feel for whether it solves my problem". Useful for prompt iteration before you wire a model into a production application, for spec-style A/B comparisons across vendors, and for confirming that a given model handles your domain (long documents, code, multilingual, etc.) before you commit to integration work.
How model routing works
Leave the model set to Auto and the OrcaRouter routing engine classifies your prompt, scores every candidate model on price, quality and live latency, and picks the cheapest model that clears the quality bar — you see the picked model on every reply. Prefer full control? Choose any specific model from the model catalog, or watch two models race the same prompt on the leaderboard's Battle Mode. Per-model pricing for everything you try here is listed on the pricing page.