# n8n

<img
  src="/brand/banner-b-03.webp"
  alt=""
  loading="lazy"
  width="1280"
  height="549"
  className="brand-band"
/>

[n8n](https://n8n.io) is a workflow automation tool with AI Agent and LLM
nodes. It reaches Hyphen through the standard **OpenAI** credential, with the
Base URL overridden.

*Setup guide: this config comes from n8n's official documentation and has not
been run end to end against the gateway. Corrections to [support@hyphen-solution.com](mailto:support@hyphen-solution.com).*

## 1. Get a key

[app.hyphen-solution.com](https://app.hyphen-solution.com) → sign in → pick a
plan or credit pack → **Dashboard → API Keys**. Copy the `sk-...` key.

## 2. Create the credential

**Credentials → Add credential → OpenAI**

- **API Key:** `sk-YOUR_KEY`
- **Base URL:** `https://api.hyphen-solution.com/v1`

The Base URL field defaults to `https://api.openai.com/v1`. Replace it. Leave
**Organization ID** empty.

Name the credential something like `Hyphen` so it is obvious in the node
picker.

## 3. Set the model on the node

Add an **OpenAI Chat Model** node (it appears as a sub-node under **AI Agent**
or **Basic LLM Chain**) and select the `Hyphen` credential.

The **Model** field has two modes. Switch it from **From List** to **ID** and
type the model name:

```
minimax-m3
```

**From List** also works. n8n populates that dropdown by calling
`GET /v1/models` on your credential's Base URL, so it shows the live Hyphen
catalog rather than OpenAI's.

:::note[Base URL lives on the credential, not the node]
Older versions of the OpenAI Chat Model node had a Base URL option in the
node's own **Options** section. Current versions hide it. If you cannot find
that field, that is why. Set it on the credential.
:::

## 4. Verify

Add a **Chat Trigger** → **AI Agent** with the OpenAI Chat Model sub-node
attached. Open the chat panel and send:

```
Reply with the single word: connected
```

Or skip the agent and use a **Basic LLM Chain** with a fixed prompt, then hit
**Test step**.

## Which model for a workflow

- **AI Agent with tools:** `minimax-m3`. Agent nodes chain tool calls and the
  flagship is built to hold a plan together.
- **Classification, extraction, routing:** `minimax-m2.5`. Same rate, and these
  jobs do not need the flagship.
- **Summarising a long document:** `minimax-text-01`. 4M context and the
  cheapest rate in the catalog.

## Troubleshooting

- **`404`**. The Base URL is missing `/v1`, or you left it on
  `api.openai.com`.
- **`401`**. Wrong key on the credential.
- **Model dropdown is empty**. The credential cannot reach the gateway. Test
  the credential from its edit screen.
- **Agent stops after one step, or returns nothing**. Reasoning models need
  output headroom. Raise the node's max tokens option. See
  [Choosing a model](/choosing-a-model#the-max_tokens-gotcha).
- **`429`**. Monthly budget spent. A workflow on a schedule can burn a cap
  fast. See [Rate limits & caps](/rate-limits).
- **Fields moved?** Check
  [n8n's OpenAI credential docs](https://docs.n8n.io/integrations/builtin/credentials/openai/)
  and the
  [OpenAI Chat Model node docs](https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.lmchatopenai/).

## Related

- [Models](/models): the full catalog.
- [Handling the 429 cap](/recipes/handling-429): scheduled workflows and budgets.