I stumbled across a neat workaround for anyone looking to use AI in their n8n workflows for learning or playing around a lot or processing low priority huge data without spending a fortune. DeepSeek (a model comparable to ChatGPT-4o) is offering huge discounts right now for processing large datasets. It’s not something they promote widely, but here’s how you can make it work:
The Discounted Pricing:
- 1 English character ≈ 0.3 token
- Input cost: $0.0000042 per character (Limited time)
- Output cost: $0.000042 per character (Limited time)
To give you an idea, processing a 500,000-character book costs just $0.042 (yep, about 4 cents). Definitely worth looking into if you’re processing large volumes of data.
How to Enable the Discount in n8n:
Go to OpenRouter and create an account.
Add a small balance (I tested it with $5).
Generate an API key.
In n8n, add the OpenAI node:
- Change the API URL to:
https://openrouter.ai/api/v1 - Use the model name:
deepseek/deepseek-chat.
Here’s the trick:
- Go to your OpenRouter dashboard.
- Block all providers except DeepSeek. This ensures you’re charged at their discounted rate.
Note:
- The pricing is insanely low right now, but it’s a limited-time deal (ends in February), so use it while it lasts.
Just wanted to share this in case it helps anyone reduce costs for their n8n workflows or other automation setups. Let me know if you’ve tried it or found any alternatives!