Claude says this conversation is too long to continue — how to carry the work on

The thread has been open since the kickoff: the statement of work near the top, the fee conversation worked through twice, the memo on its third draft. Then the reply box refuses the next message. Here is what the limit is, what Anthropic says about it, four ways to keep going, and the part none of them solve.

Nobody meets this message on a quick question. You meet it on the thread holding three weeks of a client engagement: the documents from day one, the positions you argued out, the draft you came back to each morning. Claude tells you to start a new chat, and for a thread like that a new chat is a blank page.

"This conversation is too long to continue" and "Claude hit the maximum length for this conversation": which is current

People paste three different sentences into search. Two are reported by users; the third, which Anthropic documents, has its own section below.

• The one users report now: "This conversation is too long to continue. Start a new chat, or remove some tools to free up space." Reported by users, not documented by Anthropic: a bug report about the Claude desktop app, opened 16 April 2026 on Anthropic's public GitHub (issue 48964), quotes it word for word; a second report dated 12 May 2026 (issue 58521) quotes the same sentence; and folk.com's guide "Claude 'This Conversation Is Too Long': How to Fix It" (14 August 2026) repeats it.
• The older one: "Claude hit the maximum length for this conversation." This is the wording people were quoting in 2025 (reported by users: Limited Edition Jonathan on Substack, "Ultimate Guide: Fixing 'Claude hit the maximum length for this conversation,'" 6 November 2025). That report is dated a few weeks before the November 2025 change described below, and the user reports we found from 2026 quote the newer sentence instead.

We searched Anthropic's help center on 20 September 2026 and found neither sentence there, so treat the exact wording as something that changes between app versions. Whichever you saw, the subject is the length of one conversation.

"Your message will exceed the length limit for this chat": the wording Anthropic documents

The one length message Anthropic prints in full is this: "Your message will exceed the length limit for this chat. Try attaching fewer or smaller files or starting a new conversation." It appears in the help-center article Troubleshoot Claude error messages (last updated 18 May 2026), which explains: "This error indicates that your message is too long and needs to be shortened before sending it to Claude." So this one is about the message you are trying to send, usually a large attachment or a very long paste. Anthropic's remedies are to "Break your content into smaller chunks and process them separately," to "Summarize or extract key sections before sending to Claude," or to start a new conversation.

Claude's chat length limit: what it is, and how it differs from your usage limit

Anthropic's explanation, from How do usage and length limits work? (updated 16 September 2026): "Length limits relate to Claude's context window—the amount of information Claude can work with in a single chat. Think of the context window as Claude's working memory that determines how much content it can process and remember at once."

That measures an amount of information, not a number of messages. Your questions, Claude's answers and the full text of whatever you attached all count, which is why one 80-page deck takes more room than a run of short questions. Connected tools count as well. The same article says "Tools and connectors are token-intensive" and advises: "Turn off apps you've connected when a conversation doesn't need them." That is the most likely meaning of "remove some tools to free up space" in the reported message, though Anthropic does not explain that sentence anywhere we could find.

On size, Anthropic publishes figures for paid plans only. The same article (updated 16 September 2026) says the newest models support "up to a 1M token context window, while others support 500K or 200K tokens," and its companion page, How large is the context window on paid Claude plans? (updated 1 September 2026), glosses the smallest of those as "about 500 pages of text or more." Neither page gives a figure for the free plan, so we don't either.

This is not your plan's usage limit. Anthropic draws the line in one sentence: "usage limits control how much you can use Claude across all your conversations, while length limits control how long any single conversation can become." A usage limit resets with time; a length limit belongs to this one thread. Anthropic's remedy differs too: "If you hit a length limit, you can start a new conversation or use features like projects to work with larger amounts of information more efficiently."

What Claude now does before you reach the limit

Since late 2025, Claude tries to avoid this wall by itself. Anthropic's release notes for 24 November 2025 ("Context window compaction") describe "summarizing earlier messages when a chat approaches its context limit," and its announcement the same day put it plainly: "For Claude app users, long conversations no longer hit a wall—Claude automatically summarizes earlier context as needed, so you can keep the chat going."

The help center calls this automatic context management: "When your conversation approaches the context window limit, Claude summarizes earlier messages to continue the conversation seamlessly." You may see Claude "organizing its thoughts" while it happens; what the summarizing keeps and what it lets go is covered in our guide to Claude compacting a conversation.

Our reading, an inference and not something Anthropic states: if the wall still reached you in 2026, the likeliest reasons are the conditions and exceptions Anthropic lists. None is a guaranteed fix, but check them first:

• The setting it depends on. "Code execution must be enabled for automatic context management." The switch is Settings → Capabilities → Code execution and file creation; on iPhone, tap your initials or name in the sidebar, then Capabilities. Anthropic says it is on by default on Free, Pro and Max (article updated 6 August 2026), and that on Team and Enterprise an owner can turn it off for everyone.
• Your plan. Anthropic's pages disagree here. The article updated 16 September 2026 says "For users with code execution enabled." The error-messages article (18 May 2026) and the context-window article (1 September 2026) both say "For users on paid plans with code execution enabled." If you are on the free plan, don't count on it.
• The exceptions. "Rare edge cases (such as very large first messages) may still encounter context limits." If that was your case, the remedies in the section above apply.

A note on scope: this guide covers the Claude chat apps on the web, desktop and iPhone. Claude Code, Anthropic's tool for software developers, handles long sessions its own way, and what is true there should not be assumed here.

First: the old conversation is still there

Nothing in Anthropic's documentation says a conversation is removed when it reaches the limit. Conversations sit in the left sidebar, or under Chats and tasks, which Anthropic describes as the place "to see your full chat history." You can open the thread, scroll it and copy from it. You just can't add to it.

A new chat does not load that transcript. Two features get you part of the way. On paid plans, Claude can search your past chats when you ask; Anthropic's own example prompt is "Let's continue where we left off with [project]." That is a search which pulls back the pieces it judges relevant, inside one project or across your non-project chats, never both. Claude's memory, on by default on Free, Pro and Max, "saves memory as a set of individual topics as you chat, rather than summarizing conversations after they end." Neither is the whole thread arriving in the new one, so carry the work over on purpose.

Option 1: a hand-off note (the one that usually works)

If the thread will take one more message, don't spend it on a last question. Spend it on a briefing for the next chat. Ask for decisions and exact details, not a recap of the discussion:

I'm moving this work to a new chat. Write the briefing I'll paste in as its first message. Cover, in this order: the client and the engagement in two lines; what we have decided, each with its one-line reason; every figure, name and date exactly as we last agreed it; what is unresolved and who owes what; the very next task. Name the files I gave you rather than describing them, because I will attach them again. 350 words at most.

Start a new chat, paste the briefing first, attach the files it names, and state the task in a line. If the old thread won't accept anything more, try Option 3 to get the request in, or write the note yourself from the transcript using those five headings. Note the prompt's last instruction: the documents do not travel with the note. You attach them again.

Option 2: a Project, with the documents in project knowledge

If what hurts is attaching the same files to every new chat, a Project is Claude's own answer. Anthropic: "Projects are available to all users, including those with free Claude accounts. Free users can create a maximum of five projects." Each project has a knowledge base, and "Anything you upload to this space will be used across all of your chats within that project." So the statement of work goes in once and every new chat in that project can draw on it. On paid plans, Anthropic says that when a project's files approach the limit Claude switches to looking up the relevant parts instead of loading everything, which expands what the project can hold "by up to 10x." You can also move the old conversation into the project: open the menu next to the chat's name and choose Add to project.

Two catches. First, Anthropic is explicit that the chats inside a project don't read each other: "Context is not shared across chats within a project unless the information is added into the project knowledge base." The files carry over, and Anthropic says each project keeps "its own separate memory space and dedicated project summary", so what Claude learned in earlier project chats can carry into new ones as memory. The transcript of yesterday's chat does not, so for exact figures and wording you will still write hand-off notes. Second, the project holds what you put in it, as of the day you put it there. The client's reply from Thursday and the revised deck in your inbox are not in it until you add them, and again each time they change.

Option 3: edit an earlier message to branch the conversation

Claude lets you go back and change something you already sent. Anthropic's tutorial site describes it: "Go back to any previous message, click 'Edit' to create a new conversation branch, and try different approaches. You can always return to your original version." The new branch continues from the message you edited, and what came after it stays behind on the original path.

Anthropic does not say how a branch counts toward the length limit. Users report that it helps because an edit replaces material instead of adding to it (reported by users: Limited Edition Jonathan, Substack, 6 November 2025: "You're replacing instead of accumulating"). The practical move, if the thread is refusing new messages: go back to a message from before the big attachment, edit it, and use that turn to ask for the hand-off note from Option 1. The catch: whatever came after the edit point, including any document attached later, is not in the new branch.

Option 4: start fresh, with only the files this task needs

Sometimes the cleanest continuation is a small one. Open a new chat, attach the two documents the next task depends on instead of all nine, and paste the hand-off note. Switch off connected apps the task doesn't need, which is Anthropic's own advice for saving room.

If the old thread is the only record of how a decision was reached, copy it into a document, or use Anthropic's data export: Settings → Privacy → Export data, on the website or the desktop app (Anthropic notes "it's not possible to run an export from Claude for iOS or Android"). The download link arrives by email and "will expire 24 hours after delivery." It covers the whole account, so it keeps records; it does not continue tomorrow's work.

Which fix, when

Claude's settings, labels and limits above were read on Anthropic's own pages on 20 September 2026 (sources at the foot of this guide). Anthropic changes these often; check your own account.

Where every workaround runs out

Read the catches in that table again. The note ends with "attach the documents again." The project holds the version you uploaded. The branch leaves behind what came after it. Each is the same move: you carry the client back into the chat by hand, because the chat was the only place the client's context had been assembled.

And it was never really assembled there. What matters about an engagement arrives in the email thread with the client, in the PDF they sent on Tuesday, in what you said out loud walking back from the meeting. Claude works with what you bring into it: what you type, paste, upload, or reach through an app you have connected. For most of us that means a paste, and the paste is a week old by the time the thread fills up.

With one engagement that is a chore. With seven, re-explaining the same client becomes a standing task, and the detail that didn't make it into the note is the one the client raises on the call.

A different arrangement: send it once, ask from there

What matters about a client should not have to live inside any one chat thread. Equerry is a personal assistant built on that idea: you send it things once, and you ask from there.

• CC or forward the thread. Add your Equerry address (something like mark@mail.getequerry.app) to the client thread, or forward the PDF they sent on Tuesday. Replies that keep your Equerry address on the thread keep arriving, so what it can answer from stays current without another paste. The original stays in your own mailbox. It reads the thread — not your inbox.
• Send the rest the way you already capture it. Share a PDF or a screenshot from the share sheet, dictate a voice memo after the call, or write a quick note.
• Ask about it later. "What did we agree on fees with Acme?" is a question for three weeks from now, answered from the thread you sent. When an answer draws on a note you saved, it carries a source chip: tap it to see the passage Equerry used and open the original.
• See what it kept from your notes. What Equerry remembered from a note you saved is listed; delete a note you saved and what came from it goes too.

None of this replaces Claude for drafting and thinking a problem through. Keep it, and use the hand-off note above the next time a thread fills up. The change is only in where the engagement itself is kept: with an assistant that holds what you sent it, so that the next long thread running out costs you a conversation and not the client's history.

For what Claude carries between chats by itself, see does Claude have memory. If you saw this message in ChatGPT instead, here is the ChatGPT version of this guide. If you use both, see how to use ChatGPT and Claude together, and for the wider picture, keeping your emails, PDFs, notes and voice memos in one place you can ask.

Sources, all read 20 September 2026. Anthropic help center (support.claude.com): How do usage and length limits work? (updated 16 September 2026); Troubleshoot Claude error messages (18 May 2026); How large is the context window on paid Claude plans? (1 September 2026); Release notes, entry of 24 November 2025; Create and edit files with Claude (6 August 2026); What are projects? (17 September 2026); How can I create and manage projects? (12 September 2026); Retrieval augmented generation (RAG) for projects (10 September 2026); Use Claude's chat search and memory to build on previous context (15 September 2026); Delete or rename a conversation (10 August 2026); Export your Claude data (8 July 2026). Anthropic announcement: Introducing Claude Opus 4.5 (anthropic.com, 24 November 2025). Anthropic tutorial: the Claude Academy tutorial on prototyping with artifacts (academy.claude.com, undated). Reported by users, not by Anthropic: GitHub issue 48964 (16 April 2026) and issue 58521 (12 May 2026); folk.com, "Claude 'This Conversation Is Too Long': How to Fix It" (14 August 2026); Limited Edition Jonathan on Substack, "Ultimate Guide: Fixing 'Claude hit the maximum length for this conversation'" (6 November 2025).

Last updated 2026-09-21.

All product names, logos, and brands are property of their respective owners. Equerry is not affiliated with, endorsed by, or sponsored by any other product mentioned; comparisons are provided for informational purposes.