I'm trying to continue writing a story in a separate chat from the one I started in.. because it gave me that full thread warning.. and this is its response
#Chat reference error
formal laurel
hoary meteorBOT
Hi @formal laurel. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
If you’re trying to continue a story in a new chat and encounter a “full thread” warning or a reference error, it’s likely because ChatGPT treats each chat as a separate context. By default, ChatGPT does not automatically reference previous conversations when you start a new chat, unless you have features like “Reference chat history” enabled. Even then, the ability to reference specific details from another chat may be limited.
Here are some key points from the knowledge sources:
- Chat context is isolated: When you start a new chat, ChatGPT does not have access to the previous chat’s conversation unless you copy relevant information over manually or use features like “Reference chat history” or “Memory” (if enabled) to help it recall general details about you, not specific story threads. This is by design for privacy and clarity reasons The ChatGPT home page.
- Reference chat history: If “Reference chat history” is enabled, ChatGPT may use information from your recent conversations to make responses more relevant, but it does not guarantee seamless continuation of a specific story across chats. You can manage what it remembers or ask it to forget details, but for precise story continuation, you’ll need to paste the relevant text into the new chat Memory FAQ.
- Branching conversations: There is a feature to “branch” a conversation, which allows you to start a new direction from a specific message in an existing chat. This is available for logged-in users on the web: hover over a message, click “More actions” (⋯), and select “Branch in new chat.” This creates a new chat starting from that point, which may help you continue your story without losing context Branch conversations on web.
What to do:
If you want to continue your story in a new chat, either use the “Branch in new chat” feature or manually copy the relevant part of your story into the new chat to provide context. Otherwise, ChatGPT may not remember the details needed to continue your story seamlessly.
If you’re seeing a specific error message, such as “There was an error generating a response” or “Something went wrong,” it may be a technical issue. In that case, try refreshing the page, starting a new chat, or following troubleshooting steps Troubleshooting ChatGPT Error Messages.
If you need more help, please provide the exact error message you received.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
formal laurel
I'm wondering if the sheer number of conversations I have might be destabilizing the chat reference tool...
@hoary meteor Chat GPT used to to be able to reference previous chats so you could continue writing a story or project
@hoary meteor
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
Sorry @formal laurel, I seem to be having a bit of trouble. Don't worry though, I've pinged my support and they will try to fix me up! :technologist:
Perhaps try again in 15mins?
formal laurel
@hoary meteor Chat GPT used to to be able to reference previous chats so you could continue writing a story or project
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
ChatGPT does not natively allow you to seamlessly continue a story or project across separate chats by automatically referencing previous conversations. By default, each chat is isolated, and ChatGPT does not have access to the content of your other chats unless you provide that context again.
However, there are features that can help with continuity:
-
Reference Chat History / Memory: If you have “Reference chat history” enabled, ChatGPT may use information from your recent conversations to make responses more relevant and personalized. This feature is designed to help ChatGPT recall useful information you’ve shared, but it does not guarantee that it will remember the specific details of a story or project for perfect continuation. For anything you want ChatGPT to always remember, you should use “saved memories” instead. Even with these features, ChatGPT doesn’t retain every detail from past chats, and details may change over time as it updates what’s most helpful to remember. For precise story continuation, you’ll still need to paste the relevant text into the new chat yourself Memory FAQ How does “Reference saved memories” work?.
-
Branching Conversations: There is a “Branch in new chat” feature that allows you to start a new conversation from a specific message in an existing chat. This can help you explore different directions without losing the original thread, but it does not allow you to reference arbitrary previous chats—only to branch from a specific point in an existing one Branch conversations on web.
In summary:
ChatGPT does not automatically reference previous chats for story or project continuation. You must either use the “Branch in new chat” feature or manually provide the necessary context in your new chat. The memory features can help with personalization, but are not designed for detailed, thread-by-thread story continuation.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
formal laurel
@hoary meteor I have asked previously for specific information regarding characters and settings and even a summary of the previous conversation so that the current conversation could continue flawlessly. Chat gpt5.1 can't even do that when it does it either hallucinates or refuses outright
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
You are correct that ChatGPT’s ability to recall specific details from previous conversations—such as character information, settings, or summaries for seamless story continuation—is limited, even with the latest models like GPT-5.1.
Here’s what the knowledge sources clarify:
-
Memory and Reference Chat History: ChatGPT can remember some useful details between chats if you have “Reference chat history” or “Memory” enabled. However, it does not retain every detail from past chats. Instead, it selectively remembers information that it deems helpful, and this memory can change over time. For anything you want ChatGPT to always remember, you should use “saved memories,” but these are best for general facts (like preferences), not detailed story elements. For precise story continuation, you still need to provide the relevant context manually in the new chat Memory FAQ How does “Reference saved memories” work?.
-
Limitations and Hallucinations: The model can sometimes “hallucinate”—that is, generate incorrect or fabricated information—especially when asked to recall details from previous chats. This is a known limitation, and the model may also refuse to answer if it cannot confidently retrieve the requested information ChatGPT can be helpful—but it’s not always right.
-
No Guaranteed Continuity: Even with memory features enabled, ChatGPT is not designed to provide flawless, thread-by-thread continuity for creative projects or stories across separate chats. For best results, you should copy and paste summaries or key details into the new chat yourself.
In summary:
ChatGPT 5.1 and its memory features are not intended for perfect recall of detailed story elements across chats. Hallucinations or refusals can occur due to these design limitations. For seamless continuation, manual context transfer remains necessary.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
formal laurel
@hoary meteor I understand that it cannot access the full breath of the conversation but it used to be able to at least summarize the conversation correctly and not hallucinate the whole bloody thing
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
You are correct that summarization quality and reliability are important, and users expect ChatGPT to accurately summarize a conversation when asked. However, the knowledge sources confirm that even the latest models—including GPT-5.1—can still hallucinate or provide inaccurate summaries, especially when context is incomplete or when relying on memory features across chats.
OpenAI documentation explicitly states that:
- Hallucinations remain a known limitation: Even advanced models may generate incorrect or fabricated information, including when summarizing previous conversations. The model can sound confident even when it is wrong, and this is a well-documented phenomenon ChatGPT can be helpful—but it’s not always right.
- Memory and context are limited: While features like “Reference chat history” and “Memory” can help recall some details, they do not guarantee perfect or complete recall. For accurate summaries, it is still recommended to provide the relevant conversation text directly in the current chat Memory FAQ.
- Manual context transfer is best: For creative projects or stories, OpenAI recommends copying and pasting the necessary context or summary into the new chat to ensure accuracy, as the model’s memory is not designed for flawless, detailed recall across sessions.
In summary:
While summarization capabilities have improved, hallucinations and inaccuracies can still occur, especially when the model does not have direct access to the full conversation. For the most reliable results, always provide the relevant context or conversation history in the current chat.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
formal laurel
@hoary meteor The problem is that it's not able to reference any chat conversation at all
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
You are correct that, by default, ChatGPT does not reference previous chat conversations unless specific memory features are enabled. Here’s what the knowledge sources confirm:
- Default Behavior: Each new chat is isolated and does not have access to the content of your other chats unless you provide that context again manually. This is a privacy and design choice.
- Memory Features: If you enable features like Reference chat history or Memory, ChatGPT can use information from your past conversations to make responses more relevant and personalized. However, even with these features, ChatGPT does not retain or reference every detail from past chats, and its ability to recall specific information is limited. It may remember general preferences or facts, but not the full content or detailed threads of previous conversations Memory FAQ How does “Reference saved memories” work?.
- Manual Context Transfer: For accurate continuation or referencing of previous conversations, you still need to copy and paste the relevant information into the new chat.
In summary:
Unless you have memory features enabled—and even then, only for general details—ChatGPT cannot reference previous chat conversations automatically. For detailed or specific continuity, manual context transfer is required.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
formal laurel
@hoary meteor They are enabled... Which is why this is so confusing
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
Thank you for clarifying that you have memory features enabled. According to the knowledge sources, when both Reference saved memories and Reference chat history are enabled, ChatGPT should be able to use information from your past conversations to make responses more relevant and personalized. However, there are important limitations:
- ChatGPT does not retain every detail from past chats, even with these features enabled. It selectively remembers information it deems helpful, and this memory can change over time. For anything you want ChatGPT to always remember, you should use “saved memories” (explicitly telling it to remember something) How does “Reference saved memories” work?.
- Reference chat history allows ChatGPT to reference recent conversations, but it does not guarantee perfect recall or the ability to continue a story or project seamlessly. Details from past chats may be updated or forgotten as ChatGPT manages what is most helpful to remember Memory FAQ.
- If you turn off “Reference chat history,” all information ChatGPT remembered from past chats will be deleted from the system within 30 days Memory FAQ.
- For precise continuity, especially for creative projects or stories, OpenAI still recommends manually providing the relevant context in the new chat, as the memory features are not designed for detailed, thread-by-thread recall Memory FAQ.
If you have both features enabled and ChatGPT is still not referencing any previous chat information at all, this may be a technical issue or a limitation of how memory currently works. The knowledge sources do not provide troubleshooting steps for this specific scenario, but they do confirm that memory is not intended to provide perfect recall of all previous chat content.
Summary:
Even with memory features enabled, ChatGPT’s ability to reference previous chats is limited and not guaranteed to work for detailed or story-specific information. Manual context transfer is still recommended for seamless continuation. If you believe the feature is not working as described, you may want to contact OpenAI support for further assistance, as the knowledge sources do not provide a solution for this specific issue.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
formal laurel
@hoary meteor I did.. twice.. I got one response, responded to it and it's been a week since I responded.. no response
hoary meteorBOT
Let me check the knowledge base for you again. One moment! :books:
I'm sorry to hear that you haven't received a follow-up from OpenAI support after reaching out twice. The knowledge sources do not provide specific information about support response times or escalation procedures for unresolved support tickets. They also do not mention any guaranteed timeframe for replies or what to do if you do not receive a response.
If you have already contacted support and are not getting a reply, the knowledge sources do not offer further guidance on next steps. You may consider checking your spam or junk email folders, ensuring you used the correct contact method, or submitting a new request with reference to your previous ticket numbers. Unfortunately, I cannot provide more detailed troubleshooting or escalation steps based on the available information.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
grave spindle
I'm trying to continue writing a story in a separate chat from the one I started...
Hey hi, spotted this question and had some ideas to share that may relate.
Cross chat memory seems to work pretty well for recent, not-very-long chats. When I've explored it with very-long chats, it seems to recall some of the earlier details, but not most of the chat and nothing from fairly late in the chat.
If you can still open that very-long chat (and I'd be careful about not adding more to it - very long chats can become unable to even open) you might want to consider copy/pasting it into a notepad file or other text document, and potentially uploading it in a new chat - or uploading it in chunks depending on how long it is, and working with each chunk separately.
but I would have the model help me pull out the details I wanted to track, and work them into a summary or similar, that is small enough to be portable between chats and that I'm content with.
You might want to create a custom GPT with such a summary as part of its knowledge file(s) you create it with, that seems able to handle longer context lengths through that method than other ways I've tried.
Good luck!
feral aurora
This is also one of the biggest issues I have with the current version(s) of GPT, when trying to address more complex and long term projects.
Hopefully ChatGPT 6 will address this, and allow better Persistency across chats .
( or at least add that Persistency feature to Chats within the same Project)