You can prepare settings locally, or use your own provider account. The guide does not supply a working site key while issuance is paused. Read current access and limits.
Prepare the body without sending a request
Open the request builder. For your own provider, enable the custom endpoint option and enter its documented base URL and exact model ID. Choose JSON body as the export format. Copy or download it after reviewing the task instructions.
The export is the HTTP body, not an importable n8n workflow. It contains no API key. A base such as https://api.example.test/v1 becomes the full endpoint https://api.example.test/v1/chat/completions; the example domain is not a service. Do not add the endpoint path twice.
Set up one HTTP Request node
In an existing n8n workflow, connect a Manual Trigger to an HTTP Request node. Use the request's full endpoint in this node, rather than only the API base.
| Setting | Value |
|---|---|
| Method | POST |
| URL | Your base plus /chat/completions, once |
| Authentication | Generic Credential Type, then Bearer Auth |
| Send Body | On; Body Content Type: JSON; Specify Body: Using JSON |
| JSON | The exported object, without Markdown fences |
| Response Format | JSON |
In the Bearer Auth credential, enter the key in the Bearer Token field. Do not add another Bearer prefix. If your version offers Header Auth instead, use the name Authorization and value Bearer YOUR_KEY. Replace the placeholder inside the credential, and avoid a second Authorization header in the node.
Keep the status and the model answer distinct
Turn on Include Response Headers and Status if you need the HTTP status alongside the response body. Leave Never Error off for this first check, so a non-2xx response does not appear successful. The exported body uses stream: false for a single JSON response.
For a successful node execution, inspect the HTTP body that contains choices. If n8n has wrapped it with headers and status, copy the inner body object into the response inspector, and enter the HTTP status separately. Do not paste the whole n8n execution record. For a quote-extraction task, check the assistant's inner JSON with the quote-field checker, then compare every value with the source.
Decide what should happen before a schedule runs
- Authentication or model-access failure: correct the credential or permitted model, rather than repeatedly retrying.
- A daily or concurrent limit: wait for allowance; a parallel batch can exceed the route's capacity.
- An incomplete answer or missing source fact: hold the result for review before writing it to another system.
For occasional requests, a manual step or a small script may be sufficient. A workflow tool becomes useful when you need recurring inputs, routing and managed credentials. Choose hosted or self-managed operation for that task, rather than buying a plan just to follow this page.
Updated 19 October 2026 from the linked documentation. Local request exports are checked by this site; a live n8n execution or downloadable workflow has not been certified here. Provider inference and workflow hosting follow their own terms.
Free API access and limits
Check the site free API access and limits before trying to obtain a site key. Availability follows the operator’s enablement switch and the published request limits. It is not a source of official provider keys.
ChinaGenAI is not affiliated with the providers named here. This release contains setup guidance and explicitly labeled fictional examples, not claimed live model benchmarks.