Thread author fzrunu482 described two problems: a temporary usage limit kicks in after just 5–6 messages with no images or files involved, and important chats are now getting permanently locked at "maximum length" after 2–3 weeks instead of the previous 6–7 months. In the OpenAI Community thread (50 replies, 129 likes, 4,414 views), users on Free, Plus, Business, and Enterprise all confirm the same thing. The debate isn't really about whether it's a bug — it's about where the line sits between a conversation-length limit and an ordinary usage cap.
fzrunu482 describes two distinct failures. According to him, conversations on the Free plan used to run noticeably longer; now the limit sometimes hits after just 5–6 messages. The second issue is more serious: the message "You've reached the maximum length for this conversation, but you can keep talking by starting a new chat" turns a chat into a permanently read-only artifact. Sometimes it happened right after an ordinary 3–4 line message — not after uploading files.
jotpenAI adds their own case on Plus with the GPT-5.6 Sol Chat model: "two separate new conversations in a very short time," with one hitting the max-length error after roughly half a day of use. APublish tested a different setup — Mac OS, Safari, Free plan — and saw the same pattern: the limit after 1–2 short text questions, no attachments. The message shows a specific wait time, access returns after a few hours, and then the cycle repeats.
Plus looks the same. zecanard wrote that they used to keep a single chat going for over a month, while a chat started yesterday filled up in two weeks with the same workload. coryhughes confirms: "My conversations hit the limit pretty quickly, when it used to take much longer."
corinne2 ran her own comparison test: she took a chat that had run successfully from July to September on a training program, and a rebuilt contact list that hit the maximum in under 3 days. She pasted both texts into ChatGPT, asking it to compare them and determine whether the known issue had been logged — after anonymizing the data first.
Several participants' dates line up: September 19, 2026. hidalgode on Business with a Custom GPT got the error after a single user message and a single assistant reply; the request was in Spanish, about capitalization rules. tomg on Plus started a new conversation the same day and got the same max-length message — even though the chat was only a day old.
nsschmid takes the conversation toward load: on an enterprise account, the problem reliably shows up when updating documents and using extended thinking. In his description, the chat fills its capacity, hits retry errors, and can't shut down cleanly; splitting the work into smaller tasks only helps sometimes — "9 out of 10 times the task doesn't complete."
There is a common denominator: it's not about images, video, or files. Nearly everyone responding stresses that they hit the limit on short text-only requests, and that chat lifespans have shrunk from months to days and weeks. The spread across platforms and tiers is wide — web, Safari, Free, Plus, Business, Enterprise — which makes the "one broken model" explanation look weak.
The real point of contention is different: participants are conflating two different messages. A temporary limit with a stated return time and a permanent length lock are different mechanisms, yet the thread treats them as a single regression. Nobody has posted a token count or a message tally before the limit trips, so the theories — Custom GPT, extended thinking, a specific model — remain unverified and unreconciled. There's no consensus on the cause.
Disclaimer: this is a summary of a discussion, not editor-verified facts. Opinions belong to the participants; the numbers belong to whoever reported them.
Source: community.openai.com · post in Telegram