yeah + still some models like grok4 missing info on hover and multiple issues with models docs page…
It is a bug? @condor
I saw it earlier already but will check and confirm.
@liquefy thank you for the update, will also pass it to devs
After switching briefly to Kilo Code (at least until they fix some serious memory leaks) I can tell you that 200k may not be what you think it is. In KC you can set Max Tokens (which reserves the tokens for responses, max 64k) and Max Thinking Tokens (which is reserved for thinking and must be lower than Max Tokens). Only the remaining 200k - Max Tokens is available for input / context.
This is confirmed! Was done intentionally to give you more context. We’ll be sharing more about this publicly soon.
what is the difference this and max mode?
I asked in a thread, got no response
Thanks, not aware of it, but will ask the team.