Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
There is no single fix for “Message a Model” failing in n8n: the cause may be the credential, OpenAI project, model access, request settings, network, or the particular n8n node and version. First identify whether the workflow uses the regular OpenAI action node or the OpenAI Chat Model sub-node, then test that path with a fresh API credential and a literal prompt.
First identify which OpenAI node is failing
n8n has two configurations that are easy to confuse:
- OpenAI app node: a regular workflow node that runs an operation such as “Message a Model” or the equivalent text-generation operation shown in your installed version.
- OpenAI Chat Model: a model sub-node connected to an AI Agent, Basic LLM Chain, or another compatible AI node through its language-model connection.
They do not have identical controls or necessarily use the same API behavior. Check the node’s title and connections before following troubleshooting steps. n8n’s operation names and API-mode terminology have changed across releases; the labels shown in your version may differ. See the n8n OpenAI text-operations documentation and verify the options in your installed n8n.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Run a minimal test before changing the full workflow
For the Message a Model action node
- Create a temporary workflow with a Manual Trigger followed by the OpenAI node.
- Choose the text-generation operation available in your version and select the OpenAI credential.
- Enter a literal prompt:
Reply with exactly: OK. Do not use expressions or data from earlier nodes. - Execute the node or workflow and note the exact error text and HTTP status, if shown.
For OpenAI Chat Model
- Make a temporary workflow with a Manual Trigger and Basic LLM Chain, then connect an OpenAI Chat Model to the chain’s language-model input.
- Use a fixed prompt and execute the chain.
- Compare the result with the action-node test if both configurations are available.
A minimal test distinguishes API authentication from malformed expressions, missing input, agent/tool configuration, and downstream workflow errors.
Check the API credential, project, and billing
- Open the failing node and confirm it is using the intended OpenAI credential. Use n8n’s credential test if available.
- If the key is uncertain, create a new API key in the OpenAI project intended for this workflow and replace the saved value in n8n.
- Check that the project is active, has API billing or available credits, and is permitted to use the selected model.
- If the n8n credential includes an Organization ID field, leave it blank unless you know the key requires that organization value. Do not put a project identifier in the organization field.
- Save the credential and repeat the minimal test.
n8n calls the OpenAI API; being able to sign in to ChatGPT does not by itself establish API access, billing, or model availability for an API project. OpenAI’s API quickstart describes setting up an API key and API use. API requests use Bearer-key authentication, and organization or project headers can matter for some accounts; see OpenAI’s debugging requests documentation. An n8n Community report describes a 401 resolved by removing an incorrect Organization ID, but that is a reported case, not a universal fix: community report.
Fix an empty or stuck model selector
- Save the credential, close and reopen the node, and use the model selector’s refresh control if it has one.
- Reload the n8n editor and check whether the selector populates.
- Confirm the API key belongs to the intended project and that the project can access models.
- Choose a model available to that project; do not rely on a saved selection imported from an older workflow.
- If it remains blank, test a newly created credential and run the minimal workflow. Recreating the node can help if its saved model metadata is stale.
OpenAI provides a model-listing endpoint, but availability can vary by project and account: models API reference. Refreshing is a low-cost check, not a guaranteed repair. For example, an n8n issue documented a Message a Model dropdown that remained unpopulated after credential replacement and refresh, associated with n8n 1.93.0. It was later closed without a planned fix; it does not establish that every blank selector is an n8n defect.
Match the error to the likely cause
| Error | What it usually points to | What to try |
|---|---|---|
400 Bad Request |
Invalid request body: an empty prompt, unsupported parameter or input, or an expression resolving to malformed data. | Replace expressions with literal text, use one simple user message, remove optional parameters, and execute. Add dynamic fields back one at a time while inspecting the input JSON. |
401 Unauthorized |
Invalid, revoked, mistyped, or incompletely saved API key; an incorrect organization value; or possibly a node-specific authentication issue. | Create a new key in the intended project, replace the credential, clear an unneeded Organization ID, verify the failing node uses that credential, and retest minimally. |
403 Forbidden |
The key may authenticate, but the project, organization, endpoint, or selected model may not be authorized for the requested operation. | Check project membership and restrictions, select a model available to the project, and test a basic text request. ChatGPT model access does not prove API access. |
404 Model not found |
The model ID may be mistyped, obsolete, unavailable to the project, or stale in an imported workflow. | Refresh and reselect the model, verify its ID against the models endpoint, and test a model currently available to the project. |
429 Too Many Requests or quota error |
Possible causes include missing credits or billing, a project spending limit, rate limiting, high parallel execution, or repeated large requests. | Check project billing and usage, reduce n8n concurrency and unnecessary retries, shorten prompts or output, and consider a lower-cost model. Limits and charges depend on model and account. |
| Timeout, connection closed, or other network error | n8n may be unable to reach the API, or a proxy, TLS, DNS, container, reverse-proxy timeout, or streaming issue may interrupt the request. | Check connectivity from the n8n host or container, execution logs, outbound HTTPS, proxy configuration, DNS, and reverse-proxy timeouts. If only streaming requests fail, test without streaming where the node allows it. |
A successful credential test does not rule out a malformed request body or model-access restriction. Conversely, a 401 does not prove the key itself is wrong; check the organization/project association and compare node behavior.
If OpenAI Chat Model works but Message a Model fails
If the same credential and a simple prompt work through OpenAI Chat Model but fail in the action node, the difference may be node-specific behavior, API operation, or a version regression. An n8n issue reported in March 2026 described a Message a Model authentication-header failure while Basic LLM Chain with OpenAI Chat Model worked, on n8n 1.123.25 and 2.11.4. That report does not show that those versions fail for everyone.
- Record your n8n version and the node’s displayed version or operation.
- Back up the workflow before upgrading, downgrading, or replacing the node; then retest the same minimal prompt.
- As a workaround, use Basic LLM Chain or an AI Agent with OpenAI Chat Model if that route works for your use case.
- Use HTTP Request when you need direct control of the endpoint, headers, request body, retries, or response parsing.
Using an OpenAI-compatible provider or custom base URL
“OpenAI-compatible” does not guarantee that every n8n OpenAI node works unchanged. Providers and gateways can differ in endpoint paths, headers, message format, tool calling, streaming, model IDs, and response schema. A n8n Community discussion reported that the Message a Model node did not expose a custom base URL in the same way as the model sub-node, with an n8n team member suggesting the model sub-node as an alternative. Check your installed version rather than assuming the limitation applies to every release.
- Try OpenAI Chat Model with the provider’s base URL if your version exposes that setting.
- Use the provider’s native n8n integration if available.
- Use HTTP Request for explicit control over URL, headers, request body, and response handling; this requires more setup than the specialized node.
- For local inference, use a suitable local-model integration rather than assuming the OpenAI action node supports every local gateway.
Check network access in self-hosted n8n
For self-hosted n8n, test from the same host or container environment that runs n8n, not just from your laptop. This command calls the documented models endpoint; the API key should be kept secret and never pasted into logs or support posts:
Rank #4
curl https://api.openai.com/v1/models -H "Authorization: Bearer $OPENAI_API_KEY"
If the command fails in the n8n environment, investigate container DNS, outbound HTTPS on port 443, TLS inspection, firewall rules, and HTTP_PROXY, HTTPS_PROXY, and NO_PROXY settings. If the command works but the workflow times out, inspect n8n execution logs and reverse-proxy read or execution timeouts. OpenAI documents the endpoint and authentication format in its models API reference and debugging requests guide. A Reddit report describes a local connection-closing issue; it is anecdotal, so use it as a reminder to check local networking and request configuration, not as proof of a particular cause.
Check version changes and gather useful diagnostics
OpenAI node labels, operations, and API modes change over time. A documentation mirror describes OpenAI node V2 and terminology distinguishing Chat Completions from Responses for n8n 1.117.0; confirm labels against your installed release rather than treating that wording as permanent: OpenAI text operations.
Before reporting a reproducible failure to n8n or OpenAI support, collect:
- n8n version and whether it is Cloud or self-hosted;
- node name, displayed operation, and node version if shown;
- exact error text and HTTP status;
- whether a literal prompt works, and whether OpenAI Chat Model works with the same credential;
- whether the issue occurs in the minimal test or only in the full workflow.
Do not share API keys, authorization headers, private prompts, or sensitive workflow data. Redact execution details before attaching logs.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
