When used Claude Sonnet (before GPT-5) it regularly summarized chat history. As far as I understand it reduced overall context size. Now once chat gets longer LLM responses becomes shorter and shorter. It’s not convenient to start a new chat every time and restore context in a new chat. It would be great to be able to manually ask to summarize chat history and continue the chat with a shorter context.
1 Like
I found it’s possible to do this via /summarize command in chat.