Replies: 3 comments 3 replies
-
Token usage of prompts, completions, and functions are as accurate as they can be. For example, when you make a DALL-E request, it creates an OpenAI function, and generates a completed function when it's done; both are counted towards usage in a separate module. DALL-E is specific to OpenAI. While there is an API cost, these are not technically "tokens" in the context of string values. It's the same with vision (image attachments). The balance system would benefit from counting these against the users' credits but that is not yet implemented. |
Beta Was this translation helpful? Give feedback.
-
I just wanted to add to the discussion that since this PR was merged, Vision requests are now counted towards balance usage correctly since they are technically counted as tokens, and can be counted accurately each time: #1535 |
Beta Was this translation helpful? Give feedback.
-
Hi @danny-avila I saw your previous response regarding the token usage tracking for DALL-E requests and the mention that the system would benefit from counting these API costs against user credits. Could you please confirm if this has now been implemented? Is the usage of DALL-E currently counted towards the users' credits? Thank you! |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
What is your question?
The document states that since v6.0.0, LibreChat has accurately tracked the token usage of OpenAI/Plugins endpoints, but it seems that the token usage of Dalle-3 has not been tracked
More Details
I checked the code in DALLE3.js, didn't see the any implementation。
What is the main subject of your question?
No response
Screenshots
No response
Code of Conduct
Beta Was this translation helpful? Give feedback.
All reactions