Telegram AI Agent workflow fails with "Insufficient quota detected" - looking for free alternatives for a course project

Questions
1

Telegram AI Agent workflow fails with “Insufficient quota detected” - looking for free alternatives for a course project

Describe the problem/error/question

Hello everyone,
I am building a Telegram chatbot workflow in n8n for a course project.
My workflow is:
Telegram Trigger → AI Agent → Telegram Send Message
(with Simple Memory for chat history)
The Telegram connection works correctly:
The bot receives messages
The Telegram Trigger is firing
The chat ID and message fields are available
The problem happens in the AI Agent node. I receive an error.
I checked:
The OpenAI API key is active
The API key has permissions enabled
The credential is correctly connected in n8n
The workflow reaches the AI Agent node
However, I don’t have an OpenAI billing setup (no credit card added), because this is only a course project.
According to the n8n documentation, this error can happen when:
the account has no available credits
the project has reached its usage limit
the API key is connected to the wrong organization/project
Before adding billing, I would like to ask:
Is there any way to use OpenAI models in n8n for free for testing/education purposes?
Are there any free alternatives compatible with the AI Agent node?
Would you recommend using Ollama or another local LLM provider for a beginner project?
I only need a simple Telegram chatbot that can answer messages and keep a short memory.
Thanks for any advice!

What is the error message (if any)?

Insufficient quota detected.

You exceeded your current quota, please check your plan and billing details.

Please share your workflow

Share the output returned by the last node

Information on your n8n setup

  • n8n version: Version 2.23.4
  • Database (default: SQLite): default
  • n8n EXECUTIONS_PROCESS setting (default: own, main): default
  • Running n8n via (Docker, npm, n8n cloud, desktop app): web
  • Operating system: Mac
2

Unlikely.

There is a free tier on Google AI Studio.
You can try OpenRouter as well

Why not? Just give it a try

3

The error makes sense ,OpenAI requires a paid plan even for testing and without billing set up you’ll hit that quota wall immediately regardless of how the API key is configured.

Good news is there are solid free alternatives that work perfectly with the AI Agent node.

The easiest one to get started with is (Google Gemini) Google gives you a free API key through Google AI Studio with generous limits that are more than enough for a course project. In n8n just swap the OpenAI credential for a Google Gemini credential and point the AI Agent node to it. The setup takes about 5 minutes.

Groq is another good option it’s free, fast, and supports models like LLaMA and Mixtral. Create a free account at console.groq.com, grab your API key and use the OpenAI-compatible credential in n8n pointing to Groq’s base URL.

If you want to go fully local with no API keys at all, Ollama works well but you’d need to run it on your own machine and your computer needs to be on whenever the workflow runs. For a Telegram bot that needs to be always available it’s not the most practical option unless you have a server to run it on.

For your use case I’d recommend starting with Google Gemini free tier, no credit card, works straight away with n8n’s AI Agent node and more than capable of handling a simple Telegram chatbot with memory.