it could either be a limit set somewhere in your configuration files, or the model your using can't handle the ammout of context
try running openclaw doctor --deep, them openclaw doctor --fix and restart the gateway
if its a configuration issue (that you didn't add yourself), that should fix it
if not, run /status and see how many tokens your using, and see if that exceeds the context window os the AI model your using