Playground
Test your agent, compare AI models, and verify responses before going live.
The Playground lets you test your agent before visitors see it. Ask questions, verify answers, and compare how different AI models respond.
Testing Your Agent
- Open your agent in the dashboard
- Go to the Playground tab
- Type a question in the chat input
- Review the response
If the answer isn't what you expect:
- Add more training content covering the topic
- Refine your AI instructions in Settings
- Try a different AI model
Show Sources
Enable Show Sources to see which training documents powered each response. This helps you:
- Verify the agent uses correct information
- Identify content gaps
- Debug unexpected answers
Click the sources button after any response to view the referenced documents.
Compare Models
Test the same question across multiple AI models to find the best fit:
- Go to the Compare tab (or click Compare in Playground)
- Select 2-3 models to compare
- Enter a question
- Review responses side-by-side
Available Models
| Model | Strengths | Best For |
|---|---|---|
| GPT-6 Astra | OpenAI's most capable model, advanced reasoning | Demanding tasks, complex workflows |
| GPT-6 Sol | OpenAI's new flagship-tier core model | Complex tasks, cutting-edge use cases |
| GPT-6 Luna | Fast, low-cost OpenAI model | High-volume, simple queries |
| GPT-5.6 Sol | OpenAI frontier model, advanced reasoning | Complex tasks, cutting-edge use cases |
| GPT-5.6 Terra | Strong general-purpose reasoning | General use, complex queries |
| Claude Fable 5.1 | Advanced Anthropic model for nuanced responses | Complex reasoning, detailed answers |
| Claude Opus 5.5 | Anthropic's newest Opus | Complex reasoning, detailed answers |
| Claude Opus 5 | Anthropic frontier model | Complex reasoning, detailed answers |
| Claude Sonnet 5 | Strong balance of quality and speed | General use, reliable responses |
| Claude Haiku 4.5 | Fast, cost-effective | High-volume, quick queries |
| Gemini 3.7 Flash | Fast Google model | High-volume, simple queries |
| Gemini 3.1 Pro | Advanced Google model | Complex reasoning, long-context tasks |
Model availability depends on your plan. Free plans include GPT-6 Sol (the default for new agents), GPT-5.6 Terra, GPT-6 Luna, Claude Haiku 4.5, and Gemini 3.7 Flash. All paid plans (Hobby, Standard, Pro, Enterprise) include ten core models, including GPT-6 Sol, GPT-6 Luna, and Claude Opus 5.5. GPT-6 Astra and Claude Fable 5.1 require Pro or higher.
Initial Messages
Configure what visitors see when they first open the chat:
- In the Playground, click the settings icon
- Set an Initial Message (greeting text)
- Add Suggested Messages (clickable prompts)
Preview changes in real-time before saving.
Testing Tips
- Test edge cases - Try questions outside your training content to see how the agent handles unknowns
- Test variations - Ask the same question different ways
- Test tone - Ensure responses match your brand voice
- Test on mobile - Preview how the widget looks on different screen sizes
Next Steps
- Add more training sources to improve answers
- Configure AI settings to adjust behavior
- Deploy to your website when ready