Prerequisites
1
RuAPI account and API key
Console → API Keys → create an
sk-... key. The balance tops up in USDT — see Top-up.Create the credential
1
Add an OpenAI credential
In n8n: Credentials → New → OpenAI.
2
Fill the fields
The Base URL field may sit under “Optional fields” (the gear) in the OpenAI credential form. Save the credential.
Use it in a node
1
Add an AI node
Drop an AI Agent node onto the canvas and attach an OpenAI Chat Model to it. The plain OpenAI node (Message a model) uses the Responses API in current n8n versions: RuAPI accepts those requests for non-GPT models too, but OpenAI Chat Model is the more reliable choice for an agent.
2
Pick the RuAPI credential
In the node’s Credentials field, select the credential you created.
3
Set the model
Pick the model from the list (n8n loads it from RuAPI) or enter the exact ID from the pricing page manually (“By ID” mode or an expression).
4
Turn off Use Responses API
In the OpenAI Chat Model options, turn off Use Responses API (newer node versions enable it by default). RuAPI passes Responses API requests to GPT through as is and converts them itself for Claude, DeepSeek, Qwen, Kimi and others, but n8n works with those models more reliably over regular Chat Completions.
Verify
Run the node with a simple prompt (“Hi, answer in one word”). A successful reply plus an entry in RuAPI Console → Usage Logs means the wiring works.Troubleshooting
I don't see a Base URL field
I don't see a Base URL field
Expand the optional/additional fields in the OpenAI credential form. The field is available in current self-hosted and cloud n8n versions.
My model isn't in the dropdown
My model isn't in the dropdown
Enter the model ID manually (“By ID” mode) — it must match RuAPI verbatim.
400 on Claude, DeepSeek or Qwen while GPT works
400 on Claude, DeepSeek or Qwen while GPT works
Most likely Use Responses API is on in the OpenAI Chat Model — turn it off: n8n works more reliably with non-GPT models over regular Chat Completions.