n8n's AI Agent nodes speak OpenAI's protocol. Qubax speaks OpenAI's protocol. That's the whole trick — point n8n at Qubax and every workflow can use any of 399 models, each priced up to 98% below standard retail.
Step 1: Get a key
Create a free Qubax account → API Keys → create. Top up with crypto or card.
Step 2: Add Qubax as a credential in n8n
In n8n: Credentials → New → OpenAI API:
- API Key: your
qbx-key - Base URL:
https://api.qubax.ai/v1
n8n lets you override the base URL on OpenAI credentials — that's all we change.
Step 3: Use it in AI Agent / LLM nodes
In any AI Agent, Basic LLM Chain, or LLM node:
- Select your Qubax credential
- Set the model to any Qubax model ID
Useful pattern — pick models per task, not per provider:
| Workflow task | Suggested model | Price (per 1M in/out) |
|---|---|---|
| Summarization / classification | glm-5-3-flash | $0.003 / $0.009 |
| Email drafting | gpt-5-6-luna | $0.008 / $0.048 |
| Data extraction | deepseek-v4-flash | $0.005 / $0.011 |
| Complex reasoning | claude-opus-4-8 | $2.20 / $11.02 |
| Long-context RAG | gemini-3-1-pro | $0.04 / $0.24 |
Because they're all one API, swapping the model in a node is a dropdown change — no new credentials, no new billing.
Step 4: A real example
A support-ticket triage workflow that classifies, drafts, and escalates:
- Classify (runs 1000×/day):
glm-5-3-flash→ ~$0.004/day - Draft reply (100×/day):
deepseek-v4-flash→ ~$0.01/day - Escalation summary (10×/day):
claude-opus-4-8→ ~$0.35/day
Total: ~$0.36/day, vs roughly $5+/day if every step ran on a frontier model at retail. The cheap models handle volume; the expensive model only sees the hard cases.
Tips
- Set max tokens explicitly in n8n nodes — agent loops can otherwise run longer than intended.
- Use `deepseek-v4-flash` or `glm-5-3-flash` for high-volume steps. Sub-cent pricing is where Qubax's marketplace really shines.
- Check the model page for each model's context window before feeding it 100k-token RAG payloads.
399 models, one endpoint, prices up to 98% below retail. Create your API key and wire up your first workflow.
