Skip to main content
The Chat Playground lets you try Gloo AI chat in your browser: test models, adjust settings such as temperature, and watch responses arrive before you build them into your application.
The Chat Playground is available to any organization with API access. Inference is usage-based and drawn from your prepaid credits. Messages are sent with your organization’s API key and billed like API requests: Completions (v2), the default, is a guarded endpoint priced at the model’s published rate plus 6.5%, and Responses (v1) bills the published rate. Both count against your monthly API spend limit. See Billing & Plans to get started.The Playground also has its own message allowance: 50 messages per 5 hours. A message sent to two panels in Compare Models with Sync Chats on counts as 2. When you reach the allowance, the Playground shows “Playground quota reached” and tells you how many messages remain as you get close.

Getting Started

1

Open Gloo AI Studio

Log in to Gloo AI Studio.
2

Open the Chat Playground

Click Playground in the main navigation, then select the Chat tab.
3

Connect your API key

The Playground authenticates with an API key. The first time you open it in a browser session, it shows Connect an API key. Paste an API key (it starts with sk_) into the field and click Connect. The Playground validates the key before it accepts it.Once connected, the key appears as a small chip next to the message box. Click Change to switch to a different key.Keys are kept for the current session only, so after you reload the page you need to paste a key again. If you don’t have a key, or you no longer have its value, click Create an API key to go to the API Keys page, then paste the new key. See Generate API Keys.
4

Send a message

Type your message in the text input, then press Enter or click Send to submit.
Use the quick-start action buttons (Write, Learn, Summarize, Compare, Life Stuff) to jump-start conversations with pre-configured prompts.

Selecting Models

1

Open Settings

Open the Settings panel on the right side of the Playground.
2

Open the Model dropdown

Click the Model dropdown.
3

Choose a model

Choose from the list:
  • AI Core: an auto-routing option. Gloo chooses a model for each request, so the model can change from one message to the next. The Playground currently starts with AI Core selected.
  • A specific model: models are grouped by provider (for example Anthropic and OpenAI).
  • A model family: lets Gloo pick a model from that family. Families are available on Completions (v2) only.
In the API, picking a specific model is called AI Select and picking a family is called AI Core Select; the dropdown doesn’t show those names. See model selection.
Different models have different capabilities, speeds, and costs. See Supported Models for detailed model comparisons. The model list in the screenshot below may not match the models currently offered. See the Models page for the current catalog.
Model dropdown listing AI Core, then Anthropic and OpenAI models
The Settings panel and Compare Models are available on desktop-width screens only. On a narrow screen or phone you can chat, but you can’t change settings or compare models. Use the Hide settings and Show settings controls to collapse or reopen the panel.

Choose an API version

The API selector at the top of Settings chooses which API your messages use:
  • Completions (v2): supports AI Core, model families, and denomination. This is the default.
  • Responses (v1): calls the v1 Responses endpoint (/ai/v1/responses), which is deprecated and has no guardrails or output moderation. It needs a specific model, so AI Core and model families aren’t available, and the denomination setting is disabled.
If you switch to Responses (v1) while AI Core or a model family is selected, the Playground asks you to choose a specific model. For production work, see Endpoint Types for the current guarded endpoints.

Advanced Settings

In the Settings panel, open Advanced Settings, then modify one or more settings. The first time you choose a denomination, the Playground asks for your consent to share your faith community preference with AI providers. Select None at any time to clear the preference.
Start with default settings for general testing. Adjust temperature when you need more creative responses (higher) or more consistent outputs (lower). Use max tokens to control costs in production.
After you send messages, Settings also shows a Token Usage summary of the input and output tokens used so far.
Settings panel with Model, Religious Denomination, Temperature, and Max Tokens controls, shown alongside two Compare Models panes

Comparing Models Side-by-Side

Test the same prompt across several models at once:
1

Open Compare Models

Click the Compare Models button in the top right (desktop only).
2

Pick the models

Select models for each comparison pane.
3

Check Sync Chats

Sync Chats is on by default and sends the same message to all models at once. Each synced send counts as 2 messages toward the Playground allowance.
4

Send and compare

Type your message and see how different models respond.
Compare Models view with two panes answering the same prompt, with Sync Chats on
Use model comparison when you’re:
  • Evaluating model quality for your use case
  • Testing response consistency across models
  • Choosing the right model before API implementation
  • Understanding pricing vs. performance trade-offs

Understanding Temporary Sessions

Chat Playground conversations are not saved. When you close or refresh the page, your conversation history is lost.
To preserve important content:
  • Copy valuable responses to a text editor or note-taking app
  • Take screenshots of useful examples
  • Document successful prompts for later API implementation

Transitioning to API

The advanced settings map to temperature, max_tokens and tradition; see the table in Advanced Settings. To track what your API calls use, see API Usage.

Frequently Asked Questions

The Playground is for quick testing, not storage. Requests are still processed by Gloo and the model providers. For saved conversations, implement the API in your application with your own storage.
Yes. They count toward your usage limit, are billed at standard model rates, and are also subject to the Playground’s allowance of 50 messages per 5 hours. Monitor your consumption in the API Usage dashboard.
Yes. Paste one into Connect an API key at the start of each session, and again after reloading the page. See Generate API Keys to create one.
AI Core, the Playground’s starting option, lets Gloo choose the model for each request. For your application we recommend AI Select: pick the specific model you want, so behavior stays consistent. Use model comparison to test options side-by-side. Consider factors like response quality, speed, and cost for your use case. Browse the catalog on the Models page.
Chat Playground tests conversational AI interactions. Search Playground (organizations with Data Engine access) tests semantic search across your Data Engine content library. Use Chat for general AI testing, Search for content discovery features.
Partly. You can chat from a mobile browser, but you can’t change the model, API version, or advanced settings, or compare models, because those need a desktop-width screen.

Next steps

Before you move from a quick comparison here to a change in production, run your own test set: see Compare before you switch. Pin the model you tested, as Choosing, Pinning and Updating Models explains.

Developer Quickstart

Make your first API call.

Completions API

Implement chat completions in your application

Generate API Keys

Create a key to call the API from code.

Models

Compare model pricing and capabilities.