Netlify now supports GPT-6.1 Terra in AI Gateway and Agent Runners. A Netlify Function can call the Responses API through the OpenAI SDK using the model ID gpt-6.1-terra; Netlify injects the Gateway credentials automatically. In Agent Runners, you can select the model in the model settings and run a task.
Suitable tasks: Calling GPT-6.1 Terra from Netlify Functions; running code changes or question-answering tasks with GPT-6.1 Terra from the Netlify project dashboard.
Unsuitable tasks: Inferring GPT-6.1 Terra-specific prompts, recommended reasoning effort settings, all Responses API parameters, or Agent Runner internals from this information.
Applicable model version: GPT-6.1 Terra; the Netlify Function example uses the model ID gpt-6.1-terra.
Applicable client, agent, or API: Netlify Functions + OpenAI SDK Responses API; OpenAI Codex in Netlify Agent Runners.
Recommended reasoning effort and parameters: Not specified. Agent Runners allow the model and reasoning effort to be configured per agent; the source does not specify recommended values for Terra.
The following code is from the Netlify changelog. When the function runs, Netlify AI Gateway handles the API key and authentication automatically:
import OpenAI from 'openai';
export default async () => {
const openai = new OpenAI();
const response = await openai.responses.create({
model: 'gpt-6.1-terra',
input: 'Give a concise explanation of how AI works.',
});
return Response.json(response);
};The source also says GPT-6.1 Terra is available for Scheduled Functions, Background Functions, and Edge Functions; Netlify provides caching, rate limiting, and authentication infrastructure. The announcement does not show additional configuration code for each function type.
Open the Netlify project dashboard and select OpenAI Codex under Build with an AI agent.
Open the Agent options and select Configure…, then choose GPT-6.1 Terra; set a reasoning effort if needed and save. Model and reasoning effort preferences are saved separately for each agent and apply to the current user's future runs across projects.
Enter task instructions and optional project context. Select Build to have the agent modify code and create a Deploy Preview; select Ask to answer project questions without changing code or creating a preview.
Check the run status, logs, and usage on the project's Agent runs page. When the run is complete, review the Deploy Preview before deciding whether to publish.
Netlify documentation says Agent Runners require AI features and available credits to be enabled for the team, as well as a supported team role. Check the requirements against the current team plan and account permissions before use.
flowchart TD
A[Enable Netlify AI features and confirm available credits] --> B{Choose an integration method}
B -->|Netlify Function| C[Use the OpenAI SDK and Responses API]
C --> D[Set the model to gpt-6.1-terra]
D --> E[Verify the response in the function]
B -->|Agent Runner| F[Select OpenAI Codex]
F --> G[Select GPT-6.1 Terra in Configure]
G --> H[Choose Ask or Build and submit a task]
H --> I[Check Agent Runs logs and results]
I --> J[For Build mode, review the Deploy Preview]If using AI Gateway, confirm that the project is deployed on Netlify and add the OpenAI SDK dependency according to the project's function configuration; the calling code does not need to set an AI Gateway API key or base URL.
Send a test request using the Responses API example above and confirm that the function returns a response.
If using Agent Runners, confirm that the team's AI features and credits are available, then select GPT-6.1 Terra in your personal Agent configuration.
Start with Ask mode to learn about the project or a solution. When files need to be changed, use Build mode and review the run logs and Deploy Preview.
The Netlify changelog is dated 2026-10-26 and explicitly says that OpenAI GPT-6.1 Terra is available in Netlify AI Gateway and Agent Runners.
The changelog's Responses API code sets model to gpt-6.1-terra and uses new OpenAI() without passing an API key or base URL.
The changelog says AI Gateway handles the API key and authentication automatically, and lists support for Scheduled Functions, Background Functions, and Edge Functions.
Netlify's Agent Runners documentation lists OpenAI Codex as a supported agent and says each agent's model and reasoning effort can be configured separately; the model list is provided in the Netlify interface.
Agent Runners support two run modes, Ask and Build: Ask does not change code or create a Deploy Preview; Build modifies code and creates a Deploy Preview.
The changelog does not provide GPT-6.1 Terra-specific prompts, reasoning effort options, or recommended settings; this article does not add them.
The Netlify Function code is an API call example, not a model prompt template. The English input is demonstration text from the changelog.
The changelog does not show dependency installation, function file paths, or configuration specific to each function type; follow the current Netlify Functions documentation and the project's code structure.
The models available in Agent Runner depend on the current Netlify interface and account configuration. Model and reasoning effort settings are personal preferences, not shared project-level or team-level settings.
This article summarizes official configuration and workflow guidance; it does not claim that the setup was tested in a specific Netlify project or account.
Open the Netlify changelog in the Tabbit browser and verify its title, date, model ID, Responses API code, and notes on function types. Then review the official Netlify AI Gateway and Agent Runners documentation to verify Gateway credential handling, the model configuration entry point, run modes, and preview steps. The collection date is 2026-10-27; Netlify's model catalog and interface may change later.
GPT-6.1 Terra