Algolia provides free access to OpenAI’s GPT-4.1 for testing.
Production requires your own LLM provider.
Use the same model in development and production for consistent behavior.
In the dashboard, Agent Studio can automatically detect OpenAI API keys for the global, US, and Europe endpoints.
When you add a key, Agent Studio uses the first endpoint that accepts it—https://api.openai.com/v1 (global), https://us.api.openai.com/v1, or https://eu.api.openai.com/v1.
The region comes from the key, not from your Algolia application’s region.To force a region, create the provider with the Agent Studio API and set baseUrl explicitly (the host must be an openai.com domain).
Agent Studio then uses only that endpoint, so the key must match it—a mismatch fails with 401 incorrect_hostname.European data residency requires an OpenAI project created with Europe as its region, which OpenAI limits to eligible customers.
See OpenAI’s data residency guide.
Agent Studio supports the Azure OpenAI service,
which lets you deploy OpenAI models in your Azure environment with custom compliance, security, and access controls.
Any model deployed in your Azure OpenAI resource can be used.The model selection isn’t limited to a fixed list.
You specify your Azure deployment name,
which can include custom configurations like rate limits and content filters.
Requests use the region of your configured Azure OpenAI resource.For more information, see Azure’s OpenAI documentation.
Agent Studio supports Google Gemini models using the Generative AI API.
gemini-2.5-flash
gemini-2.5-flash-lite
gemini-2.5-pro
gemini-3-flash-preview
gemini-3.1-pro-preview
gemini-3.1-flash-lite
gemini-3.5-flash
If Google deprecates a Gemini preview model,
agents using that model are automatically migrated to the equivalent generally available model and continue working without changes.
Regional availability and data residency depend on the provider or deployment behind your base URL.
Check your provider’s documentation for supported regions.
Go to Agents in the Agent Studio dashboard.
Create a new agent or select an existing one.
For an existing agent, change the provider and model under Provider and model in the agent’s details view.
Changes take effect immediately.
To update or delete providers,
open Agent Studio’s Settings page and select the provider’s View actions menu.
Provider updates affect all agents using that provider.
If you delete a provider that’s in use, its agents stop working until you assign a different provider.
Different models support different configuration parameters.
Agent Studio automatically detects and applies appropriate settings based on the model you select.
To configure these parameters, use the update agent API method.
Temperature support
Reasoning models
Most models support temperature configuration (0.0 to 2.0) to control randomness in responses.
Use:
Lower values (0.0 to 0.5) for more deterministic, focused responses.
Higher values (1.0 to 2.0) for more creative and varied responses.
By default, Agent Studio doesn’t apply a temperature value.
Models use their provider’s default (typically 1.0).You can set the temperature in your agent configuration:
{ "config": { "temperature": 0.7 }}
Models that support temperature:
All Anthropic Claude models
All Google Gemini models
Most OpenAI GPT models (except GPT-5 and o-series)
Most OpenAI-compatible models
Models that don’t support temperature:
GPT-5 series models
o-series models (o1, o3, o4)
When temperature isn’t supported, it’s automatically excluded from requests.
Some models support advanced reasoning capabilities with different configuration options depending on the provider and model series.
Agent Studio supports these Gemini configuration parameters:
For Gemini 3.x, thinking_level accepts "low", "medium", or "high" (default). Reasoning can’t be turned off, and Agent Studio ignores thinking_budget.
For Gemini 2.5, thinking_budget accepts -1 for automatic reasoning (default), 0 to turn off reasoning, or 512 to 24576 for a fixed token budget.
For example, configure Gemini 3.x:
{ "config": { "thinking_level": "high" }}
Configure Gemini 2.5:
{ "config": { "thinking_budget": 2048 }}
Invalid values return errors.
Higher reasoning effort increases latency and token usage.
For complete parameter specifications, see the Gemini API documentation.
