Experiences
Give an ongoing conversation clear context
Your conversation thread, your user persona and memory slots have different roles. The thread holds the exchange, the persona describes your part and Premium Plus has a separate memory allowance. Restate important current facts when continuing a story instead of expecting automatic, complete recall.
Topics in this article
Resume with the facts that matter now
After a break, identify the participants, location and decision still open. Correct inaccurate facts before introducing actions that rely on them. For example, explain that your characters have reached the gallery but have not chosen an exhibition yet.
Provide a consistent account of the active scene rather than incompatible summaries. Specify where you want to resume and which earlier matters are resolved. Keep the recap current so the companion is not given competing versions of the event you intend to continue.
Separate alternate continuations with Clone Chat
Where available, Clone Chat copies the conversation history up to the branching point. Continue with another choice in that copy while leaving the original story separate. Later messages develop within their own threads, rather than keeping the clone and original synchronised as the same conversation.
The listed Premium Plus allowance is twenty memory slots and five cloned chats. These counts describe account allowances, not a word limit or context-window capacity. They do not promise that every previous detail will remain available or be recalled accurately in future replies.
If you cannot find a supported memory or cloning option, confirm you are signed into the account with Premium Plus. Contact our team with your plan, browser and the missing control's name. A password or a complete private chat is not needed to describe that access issue.
Use a persona for the part you play
A supported user persona introduces your own role and background across exchanges with different companions. It describes your side of the fiction, not the personality of the character responding. Editing a persona does not replace that companion's definition or redesign its traits.
Include only details you intend to bring into other conversations. An invented background can explain your role without disclosing your home address, password or other sensitive information. Personal data you need to keep private does not belong in a reusable fictional introduction.
Consider the consequences of reset and deletion
Retain a thread you may want to revisit. Reset Chat begins again, and Delete Chat removes the conversation without restoration through our normal chat controls. Neither is merely a way to hide older messages while leaving the interaction and its history unchanged.
Memory slots count entries, not the amount of conversation considered for a particular reply. Keep a concise recap of current essentials and resolve contradictions before advancing the story. An allowance count alone cannot tell you which earlier details a response will use.
Your questions
Can I rely on complete recall throughout a long conversation?
We do not guarantee permanent or flawless recall. Repeat essential details and keep the present scene consistent. Supported memory options and your user persona may provide context where offered, but neither establishes that every previous message will be remembered.
Can the regular chat controls recover a deleted thread?
A thread removed using Delete Chat cannot be brought back through our normal conversation controls. Leave the conversation in place if you may need to return to its history rather than deleting it to tidy the view.
Do twenty memory slots specify the model's context-window size?
No. They are an account allowance, not a published model capacity in words or tokens. To resume a scene, give a brief recap such as: 'We are adult colleagues in the archive. The parcel has arrived, but we have not opened it.' That establishes the current situation without assuming that every earlier detail will be recalled.
Do V2 or memory allowances tell me which chat model I use?
Neither identifies the language model. V2 names a character-builder version, and memory counts describe allowances. We do not specify a universal model or fixed context-window size for every chat, so the available account options should not be interpreted as access to a particular model.