Why are my n8n Assistant credits running out quickly?
In this article
In this article group
- Can I have API access while I'm on a trial?
- Troubleshooting Common Login Issues
- Migrating from cloud to self-hosted
- I'm interested in the Enterprise Plan, how can I talk to a sales representative
- I'm looking to become an n8n Ambassador
- Contact the n8n Events Team' or 'Events Sponsorship and Participation Enquiries
- Which license do I need for my use case?
- I submitted a node
- Accessing the n8n Assistant
- Request for SOC-2 report
- How to delete a workflow
- I'm interested in a product demo. How can I schedule this?
- How to Retrieve Your n8n Level 1 or Level 2 Course Results
- Becoming an n8n template creator and getting help
- Understanding Workflow Publishing and Unpublishing n8n 2.0
- Work Experience and Internship Requests at n8n
- I have a marketing collaboration, sponsorship, or press inquiry
- How do n8n Academy and Community Forum accounts relate?
- Can I add or reset my Gateway credits?
- Do you have an Affiliate program?
- I want to partner or collaborate with n8n
- My n8n Community account was silenced. What should I do?
n8n Assistant credits are used when the Assistant processes your requests. The number of credits used depends on the work needed to complete a request, not just on the length of your message.
What happens when I send a request?
n8n Assistant is an agent that can perform several steps to help with your task. For example, when you ask it to troubleshoot a workflow, it may read the workflow, inspect an error, look up information, suggest changes, and check the result.
While working on one request, the Assistant may call the AI model multiple times. Each call processes information, such as your message, workflow details, conversation context, and results from earlier steps.
For example, “Why does this node fail?” is a short question. But answering it may require several steps before the Assistant can suggest a fix.
How are n8n Assistant credits calculated and deducted?
n8n Assistant uses credits for the work processed while handling your request. The underlying AI model processes text in units called tokens.
Credit usage depends on the number and type of tokens processed, as well as the model used. Processing can include:
Reading input, such as your message and workflow context
Generating output, such as the Assistant's response
Writing information to the cache
Reading information from the cache
These types of processing don't necessarily use credits at the same rate. A single request can involve multiple model calls, so there isn't a fixed number of credits per message.
We aim to use advanced AI models that provide high-quality results when building and troubleshooting workflows. The model used can also affect how many credits a request consumes.
The credits used are deducted from your instance's shared Assistant credit balance.
Why can a long conversation use more credits?
As you continue a conversation, the context available to the Assistant can grow. This context may include your earlier messages, workflow details, and results from previous steps.
The Assistant uses a cache to store information that can be reused during the conversation:
A cache write stores information for later use.
A cache read retrieves information that was stored earlier.
Both use credits. Writing information to the cache can cost more than reading the same amount from it, but caching lets the Assistant reuse information in later model calls.
A long conversation can require more context to be stored and read. If a request involves several model calls, context may be processed multiple times. This is one reason a short follow-up can use more credits than you expect.
How can I reduce credit usage?
For recommendations on making your credits last longer, see Credit usage in the n8n documentation.
How can I check my credit usage?
To see your Assistant credit balance and usage, follow the steps in Track Assistant credit spend.
What if my credit balance doesn't look right?
If you see conflicting balance messages or believe credits were deducted incorrectly, contact n8n Support. Include the approximate date, time and time zone, and a screenshot if available. This will help the team investigate.