Skip to main content
Playground with the chat list in the sidebar and an open conversation about model context windows

Using the Playground

Playground gives teams one interface to chat with every model and agent in a workspace. No separate subscriptions or accounts needed.
  • Every provider, one place: OpenAI, Anthropic, Google, Mistral, custom models, all in a single chat.
  • Switch models mid-conversation: compare outputs across providers without leaving the thread.
  • Chat with agents: deploy agents built in Orq.ai directly into Playground for the whole team. See Build Agents to learn how to enable agents for Playground.
  • Admin-controlled access: scope available models and agents per workspace.
  • Files built in: upload images, PDFs, and documents directly into the conversation.

​
Starting a Chat

  1. Click the Playground toggle at the top of the sidebar.
  2. Click New Chat in the sidebar to start a fresh conversation.
  3. Select a model or agent using the selector in the input bar.
  4. Type a message in the input box at the bottom.
  5. Press Send or hit Enter to submit.
Past conversations appear in the left sidebar, named automatically based on the content of the exchange.
Hover over any chat name and click to rename or delete it.

​
Models and Agents

The input bar includes a selector that lists all models and agents enabled by the workspace admin. Select the desired model or agent before sending a message. Models can be switched mid-conversation.
  • Models are sourced from the AI Gateway, giving access to models across all configured providers.
  • Agents are built in AI Studio around specific use cases and can use tools, knowledge bases, and multi-step reasoning. Playground is a great way to test and validate agents before rolling them out to the team.

​
Input Options

Click in the input bar to open the options menu. Three options are available depending on the selected model or agent:
Agents are pre-configured with their own model, tools, and settings. Thinking and Web Search cannot be toggled from the chat.

​
Upload Photos & Files

Attach images, PDFs, and documents directly to a message. Maximum file size is 20 MB. Available for any model with vision or document capabilities.
Only available for models that support Web Search.
Toggle web search on or off. When on, the model retrieves up-to-date information from the web to ground its responses.

​
Thinking

Only available for models that support Thinking.
When Thinking is enabled, the model reasons through the problem before writing its response. This improves accuracy on complex or multi-step tasks at the cost of additional tokens and latency. Selecting Thinking opens a submenu with two modes:
  • Standard: the model applies a moderate level of reasoning.
  • Extended: the model reasons more deeply, best suited for difficult problems that benefit from more thorough analysis.