Information on ctx

[Alexmen97]

Hi, I'd like to know what I should put in the CTX. 256k? 190k? 1M? How does OpenClaw work? Does it automatically compact even if the model doesn't support that CTX? There's nothing written about this in the documentation. I'm waiting for a reply. Sorry for my English.

[brunobuddy]

Hey @Alexmen97 right now we force 2M context but we are thinking about making it dynamic so we can compact ctx when needed. Would a feature like that make sense for you?