|can you help me?
Why can’t i use more token in gemini? gemini flash 2 model has 1m token. isnt it?
15:15:15 [vite] Internal server error: Unable to submit request because it has a maxOutputTokens value of 655360 but the supported range is from 1 (inclusive) to 8193 (exclusive). Update the value and try again.
I wondered as well today as I saw this in another topic/issue.
As far as I see and also find in forums, the output token limit is 8192 for the most models:
the 1m you mention is just for Input, not Output
Alright, but here’s the issue: I think it sends all the code as input and gets the more code in return. However, a medium-scale project could be done by outputting only the parts that need to be added or modified. The token limit runs out quickly. While the input supports 1M or 2M tokens, the output unfortunately supports much less. If I recall correctly, apps like Cursor directly provide only the added or changed sections in the output. I had tried Bolt News before, and I remember it generated longer code.
You are correct. Bolt.diy sents the whole project every prompt at the moment and the DIFF Implementation is already on the Roadmap:
