Copilot exceeds context limit

[jinzhongjia]

I sometimes refer to multiple files, which is the copilot that reports up and down issues beyond the limit, but it doesn't seem to happen in vscode

Is there any good way to solve it

[olimorris]

I believe there's a 90k token limit in copilot

[jinzhongjia]

Yes, is there any way to avoid the problem of exceeding the limit, or we can automatically pass files in batches to avoid manual batches

[olimorris]

I wouldn't have thought so. The token calculation is done on the GitHub end.

[olimorris]

As a word of advice, if you're putting over 90k of input tokens into your prompt, you may want to look at VectorCode

[jinzhongjia]

OK, it seems that this is only possible through vectorcode
Copilot can use gemini 2.5 pro but does not open up a higher context. It's a pity