Skip to main content
This quickstart helps you run a single prompt, compare models, and save a reusable chat session.

​
Step 1: Open Playground

  1. Sign in to Bud Admin.
  2. Select your project/workspace context.
  3. Click Playground from the left navigation.

​
Step 2: Start a Single-Model Chat

  1. Choose a model/endpoint in the selector.
  2. Enter a prompt (example: “Summarize this support ticket in 3 bullet points”).
  3. Click Send.
  4. Observe output quality and latency.

​
Step 3: Tune Generation Settings

Open Settings and modify:
  • Temperature
  • Response length/token limit
  • Repeat penalty
  • Stop strings
Run the same prompt again to compare behavior changes.

​
Step 4: Compare Models

  1. Open a new chat window/pane.
  2. Select a second model.
  3. Reuse the same prompt in both panes.
  4. Compare response quality, consistency, and speed.

​
Step 5: Save and Revisit

  1. Keep the session in chat history for later review.
  2. Use message-level controls (copy/edit/retry/feedback) to iterate.
Use the same prompt template across models to make comparisons fair and repeatable.

​
What to Do Next

  • Move to Playground Concepts to understand session architecture.
  • Follow Creating Your First Playground Workflow for a full evaluation loop.