Playground Views

Save and revisit Playground sessions with Views

A Playground View is a named snapshot of your Prompt Playground session. It stores your model and parameters, messages and tools, connected datasets or spans, and generated results—so you can pick up where you left off, share a setup, or compare runs without rebuilding from scratch.

Save and name a view

Start from New Playground, name your view, and work as usual; your session is tied to that view. To branch without overwriting, click Save as New View and give the copy a clear name.

What gets saved

When you save a view, the following are included:

LLM configuration

• Provider (OpenAI, Azure OpenAI, Vertex AI, AWS Bedrock, and others)
• Model and parameters (temperature, max_tokens, top_p, and related settings)
• Custom endpoint settings, when applicable

Prompt setup

• Full conversation (system, user, assistant, tool roles), content, and attached images
• Tool and function-calling configuration

Generated results

• Model responses from the session
• Latency, token usage, and evaluator scores when applied

Data context

• Connected dataset or span replay source

Auto-save

Views auto-save as you edit; an indicator shows when a save is in progress. You can turn auto-save off in Settings.

Manage views

Manage views in your playground listing page.

Next up

You are at the end of the Improve Prompts path. For the full hub story, see Save and version prompts; to recap concepts, see What are prompts?.