
I think I’m done with Raycast - hidden token costs?
I noticed an increase in my OpenAI API spending and compared an older and a newer version of Raycast.
I used the same custom AI Chat preset with identical instructions and submitted the same text three times in each version. Each version had a fresh API key used only for this comparison.
The OpenAI dashboard showed:
- Older version: $0.02
- Newer version: $0.13
That’s approximately 6.5× higher spending, based on the displayed amounts.
Raycast’s documentation mentions server-side prompt processing and additional context in AI Chat. However, I haven’t found an explanation for this difference in a simple, repeated task.
I don’t know the cause yet. Can the Raycast team explain what might account for the increase, and how users can inspect the full token usage of a request?
Has anyone else noticed something similar? Unexpected API costs like this make me reconsider using Raycast.