A token is a unit of computation used by LLMs (Large Language Models). When an LLM generates a response, it consumes a certain number of tokens. When an LLM is used to embed knowledge items, it also consumes a certain number of tokens. Here’s a rough estimate of what one token is equivalent to:
An LLM token typically represents about 4 characters or 3/4 of an average English word.
As a rough estimate, 1 word ≈ 1.33 tokens.
Those estimates are for English text. For other languages, the token to word ratio is usually higher.
Knowledge tokens are consumed when you upload new knowledge items to Droxy.For example, if you upload a company guidelines PDF with 10,000 words, it would use approximately 15,000 knowledge tokens.
Tokens are counted per uploaded knowledge item.
There are no limits to the number of knowledge items you can upload.
Deleting a knowledge item will not reduce your token usage.
We monitor the storage used by all knowledge items uploaded to your account. That storage usage is reset after every upload or deletion of a knowledge item.
If you need higher limits that are not available on any of our existing plans, please contact us at contact@droxy.ai. We’re happy to work with you to find a custom solution that fits your specific needs.
When you reach a usage limit, the feature associated with that limit will no longer work and you’ll receive a notification. Here’s what happens for each type of limit:
Message Generation: When you hit the generated tokens limit, your agents will not be able to respond to messages
Knowledge Base: When you reach the knowledge tokens limit, you will not be able to upload more knowledge items
Voice Calls: When you hit the call minutes limit, you will not be able to start new calls
In all cases, you can easily upgrade your plan to increase the limits. If you don’t see limits that are suitable for your business, reach out directly to us at contact@droxy.ai