Chat & Playground

The console has two places to call models without writing code: Playground and Chat.

Playground

Playground in the left navigation is the console's own debugging surface, available as soon as you sign in.

• Pick a model and tune parameters (temperature, max_tokens, and so on)
• Set a system prompt
• Switch between streaming and non-streaming
• See the response and what the call cost

It is the quickest way to try a model, compare output, or track down a parameter problem.

Chat

Chat in the left navigation is not a second built-in chat box — it is a list of third-party clients. Picking one hands it the RouterBus address and an API key, so you do not have to enter the configuration by hand.

The list includes clients such as Cherry Studio, DeepChat, LobeChat, OpenCat and AI as Workspace, plus CC-Switch. Desktop clients are launched through their own URL scheme; web clients open in the browser.

You can also choose Chat on a specific key's row on the API Keys page, which hands that key to the client so the usage is attributed to it.

Which key to hand over

Keep the key you give to third-party clients separate from the one your own code uses. Then revoking one does not disturb the other.

Billing

Both bill exactly like the API: tokens come out of your balance and the calls land in Usage Logs. Prices are in Model Square.