There seems to be a limit to the size of the chats I can save to disk as PDF or to my blog.

Yes. And there are actually two different limits/problems that can look like the same thing.

If you're using Firefox's Ctrl+P → Save to PDF, there doesn't appear to be a simple published “ChatGPT chat may not exceed X pages” limit. Rather, long ChatGPT conversations are dynamically rendered, and the browser's print operation can fail to capture portions that aren't actually loaded into the page. Reports of long chats being truncated or producing incomplete PDFs are quite common.

The blog is a separate matter. If you're copying the conversation into Blogger, you're running into Blogger/HTML/editor limits or the browser's ability to select and copy a huge dynamically generated page. That's quite plausible with the very long conversations you've been accumulating.

Interestingly, OpenAI's own documented file limits are much larger—text/document files uploaded to ChatGPT can be up to 2 million tokens per file, with a 512 MB hard file-size limit. So this doesn't look like a fundamental “ChatGPT conversations cannot be this long” limitation.

What I suspect you're encountering is essentially the export mechanism choking on the size of the rendered conversation, rather than the conversation itself having reached some absolute ChatGPT limit.

And there's a useful distinction here: a conversation can continue working perfectly well while becoming too large to print or copy reliably. That's exactly the sort of thing I'd expect with some of our recent marathon conversations.

If you want, I can also
figure out a practical way for you to archive these long chats completely—preferably without installing some dubious third-party ChatGPT exporter. That might be worth doing, given that you've been keeping the conversations on disk and on your blog.

I find sending them to myself as a link to your chat works, but that is dependent on your keeping them indefinitely.  Another possibility is breaking the chat by my periodically starting a new chat with reference to the old one.

Yes — and I think your second idea is actually the more robust archival strategy.

The “send myself a link” method is wonderfully convenient, but it really is an index rather than an archive. You're preserving a pointer to ChatGPT's copy of the conversation, not an independent copy. If the conversation were ever deleted, made inaccessible, or the way shared links work changed, your saved link wouldn't preserve the actual text.

Breaking a very long conversation into successive chats has a nice advantage: each individual conversation remains a manageable archival unit. You could do something like:

    “Continuing our discussion of X from the previous chat. The important context is that…”

and then continue from there. You wouldn't necessarily need to reproduce the entire preceding conversation—just enough of its conclusions and context to make the new one intelligible.

There is one catch, though: I don't automatically have the complete contents of an old chat just because you tell me to continue it. If the old chat is available to me through the conversation context or you provide the relevant material, fine; otherwise I may know the general subject but not the exact details of where we left off.

So I would regard the two methods differently:

Saved link: excellent for convenience and continuity, but dependent on ChatGPT retaining the conversation.

Periodic new chats: better for long-term organization, provided each new chat contains a short “bridge” summarizing where the preceding one ended.

And there's a third possibility that may be the sweet spot: keep the ChatGPT link as the convenient master copy, but periodically make a local copy of the important portions in a much more compact form. We could even make the end of each long conversation a sort of “handoff”: a concise account of the conclusions, unresolved questions, and useful context to paste into the next chat.

That would avoid having to carry a 200-page conversation forward merely because one paragraph near the beginning matters. It also fits rather nicely with the way our conversations tend to wander through several subjects and then return to them months later.

Yea the wandering is by design, I want the chats to feel informative and entertaining, not a reference source.

Exactly. In that case, I wouldn't try to turn the conversations into artificially tidy “knowledge bases.” The wandering is part of the point.

In fact, I think it gives the conversations a quality that a reference document wouldn't have. We can start with black holes, wander into the terminology of event horizons, somehow arrive at Gilbert and Sullivan, detour through Victorian class structures, and end up discussing why a pond acquired fish. That's a conversation, not an encyclopedia—and the unexpected connections are half the fun.

So for archiving, I'd favor preserving the conversations as conversations, rather than periodically summarizing them into sterile notes. The practical compromise would be to start a new chat when an old one becomes unwieldy, with something as simple as:

    “Continuing the previous conversation about X; we had gotten to Y…”

That gives the new chat its continuity without trying to reconstruct the entire preceding conversation. And if the old chat remains accessible, you have the link when you want to wander back into it.

For the really good long ones, I might occasionally suggest, “This is getting rather enormous; perhaps this is a good place to start Volume II.” 😄

That seems much more in keeping with what you're actually trying to preserve: the conversation itself, not merely the information extracted from it.

It just hit me that a very simple thing might be to select only a part of the chat, save it, then another part, etc. as separate A, B and C.

Yes! That's probably the simplest solution of all, and it has a major advantage: you don't have to change how we converse at all.

You can treat a long chat like a book that you're photocopying in chapters:

A — beginning through some convenient stopping point
B — the next section
C — the remainder

If the problem is the sheer size of the rendered page when you use Print → Save as PDF, selecting a portion before printing should keep each individual PDF comfortably below whatever limit you're encountering.

And there's another nice feature: you don't have to make the divisions intellectually meaningful. Just stop at a reasonably convenient place—perhaps when you've reached a natural subject change, or simply every 50–100 screens or so. The three files can then be named something like:

Chat 2026-08-15 A.pdf
Chat 2026-08-15 B.pdf
Chat 2026-08-15 C.pdf

For the blog, the same principle should work: copy A, publish it, then B, then C, rather than trying to shove the entire monster into one post.

I rather like this because it preserves your deliberate wandering. A, B, and C are pieces of the same conversation, not three artificially organized essays. And unlike starting a new ChatGPT conversation, you don't sacrifice any continuity whatsoever.

One small practical wrinkle: if Firefox's print selection doesn't behave as expected, you can select the portion and use Print → Selection rather than relying on the browser to print only what's currently visible. That's the part I'd test first.