Обновления

OpenAI just released GPT-6 Astra, their newest flagship model, and we've added it to Sudowrite.

OpenAI is calling Astra "a new generation of intelligence"—their most capable model yet. Their launch has focused on how GPT-6 enables complex, technical work, as well as its incredible instruction-following capabilities. For authors, that can mean much stronger continuity across chapters or tighter adherence to the Story Bible. It also makes Astra a great candidate for building and running multistage Plugins on Sudowrite.

As for prose quality, our early tests are promising. It sounds a bit more human than prior GPTs. But we still have a lot to learn about how well it handles long-form fiction—we’re excited to hear what you think of it.

GPT-6 Astra is available now in Write, Draft, and Plugins.

Anthropic just released Claude Fable 5.1, an upgrade to their flagship Fable model. We've added it to Sudowrite.

Fable 5 has been a recent favorite among Sudowrite authors, and 5.1 is a refinement of that same flagship. Anthropic points to performance gains and (notably for authors) fewer false refusals. In plain terms: 5.1 should be even less likely to balk at legitimate story content—dark themes, violence, heat—so your prose can go wherever the story needs to go.

In testing, it was nominally cheaper to use than Fable in the Write button, whereas Draft credit usage is basically identical. We haven't put it through its paces on long-form fiction just yet, so we can’t fully attest to its prose. Maybe you’ll let us know what you think?

Claude Fable 5.1 is available now in Write, Draft, and Plugins.

We just added a batch of formatting improvements that will give you more control over paragraph spacing and make copy/paste more reliable.

Set your own paragraph spacing

There’s a new Paragraph Spacing setting in your account settings (under the gear ⚙️ icon in the upper right). Choose your preferred paragraph spacing, and it applies consistently across your projects—no more hand-inserting blank lines for readability.

Tidy up empty lines in one click

If you (or a prior import) left empty “spacer” lines to provide a visual buffer, a new toolbar button will collapse them for you—either at the document or project level—so you can use the new paragraph spacing settings instead.

Copy and paste that preserves your formatting

Pasting into Sudowrite from Word, Google Docs, ChatGPT, or Claude now preserves formatting far more reliably. And copying out of Sudowrite will no longer double your line breaks. Imports come in cleaner too, without phantom empty paragraphs.

A more polished editor all around

Headings now scale with your text size and match your chosen font. Nested bullets now use proper markers with tighter spacing. And your formatting—bullets included—should now appear in the editor exactly the same as it does in cards in your History section.

Big thanks to everyone in the community who flagged these copy/paste and spacing issues in recent months. This one’s for you.

The team has been hard at work on fixes and key improvements this week. Here’s what’s new.

Draft won’t charge you more than its estimate - We improved the Draft tool’s estimates recently, and in doing so made that estimated range into a promise. You’ll no longer see credit charges in excess of the estimate’s ceiling in the Draft planning step. Hooray!

One note: While making this update earlier this week, we briefly broke the retired My Voice model options in Draft. My Voice users hit a credit-estimation error and couldn’t generate. The issue lasted about a day, but it’s fully fixed now. Sorry about that!

Chat Improvements

Chat reads and inserts more reliably - We fixed a batch of issues that made it difficult for Chat to read your document or insert its edits. Reading and inserting are more reliable now.

Failed plans shouldn’t use your credits - It was possible for a complex multi-step plan like a complete manuscript review to fail and still charge you credits. We fixed that, and you won’t be charged in those cases.

Reduced internal jargon in your messages - Chat’s own reasoning sometimes includes noisy jargon—document IDs, “tool” references, etc. We tweaked it so it would stop referencing that in conversation.

Other Improvements & Fixes

Write now completes the sentence - The Write button previously prioritized your word count preference over completeness of thought—meaning it could stop mid-sentence. It now runs to the nearest complete sentence, even if that means a few extra words.

Custom Rewrite instructions stick around - The Rewrite button’s “Customize” instructions had quietly become single-use. We made it work like it used to, so they stay put for repeated rewrites.

The highlight toolbar waits to appear - When you select text, the pop-up toolbar now appears after you release the mouse button instead of mid-drag—so it stops obscuring the lines around your selection.

Find and Replace now handles doc switching - We changed Find and Replace so that if you have a term entered when switching documents, it re-runs automatically in the new doc.

New titles stay put - Typing a title too quickly in a brand-new project could sometimes snap the title back to “Untitled.” Your title now sticks.

A cleaner document list - We improved the document list in your left bar—giving a bit more space, improving contrast for accessibility purposes, and making (•••) menus and toggles only show up when you need them.

Undo no longer undoes more than it should - We fixed a bug with undo where it could reverse the past few edits rather than just your most recent edit. Meanwhile, the editor’s Undo button now works on the first click after you insert text from a card. (Did you know there’s an Undo button in the editor toolbar? I didn’t, and I write these!)

It’s also worth calling out a noteworthy fix on the Sudowrite mobile app this week:

We fixed the cursor jump in Story Bible - On mobile, the cursor was jumping to the end of certain Story Bible fields after each character, making them frustratingly difficult to edit. We fixed that!

Some fresh improvements to report: a nice price drop on a couple of our OpenAI models, and another round of Chat improvements based on your feedback.

💸 GPT‑5.6 models just got cheaper

OpenAI recently lowered its prices on some of its models, and we're passing the savings on to you. GPT‑5.6 Luna now uses 70% fewer credits on Sudowrite, while GPT‑5.6 Terra uses 10% fewer. We tweaked GPT‑5.5 while we were at it, and it now uses about 12% fewer credits. You don’t have to change anything to benefit from this—if you write with any of these AI models in Write, Draft, or Plugins, your credits will just go further now.

💬 Chat fixes & improvements

We remain hard at work, making Chat behave the way you expect:

• See Chat's thought process in Chat Only — The “working” box now shows up in Chat Only mode, just like in Allow Edits.

• Narrow questions stay narrow — Asking about one specific thing no longer kicks off a full Feedback sweep across every review type, so you won't burn credits you didn't mean to spend.

• Resend actually resends — If a response stalled, hitting Resend could sometimes throw an error instead of retrying. It now consistently retries.

• More accurate document edits — We made it so when Chat inserts changes into your manuscript, it places them more precisely.

• Fewer false "stuck in a loop" messages — Chat would occasionally claim it was stuck in a loop even after finishing the task. We've cut those way down.

• Story Bible shortcuts in your current chat — Hitting a generate button in Story Bible no longer creates a brand-new chat thread. We made it so the prompt box now pops into your current chat.

🛠️ Other improvements & fixes

• Generate Scenes costs less on long projects — We’ve optimized the Generate Scenes feature. It now pulls less redundant context, which reduces the cost of scene generation on longer projects.

• Clearer Draft limit warning — We reworded the Draft input‑limit warning—occasionally triggered in super large projects—so it's less alarming and easier to address.

• Resizable comment column — We changed the comment column so that it’s fully resizable. Just drag it to your preferred width and it’ll stay that way.

Keep the feedback coming folks! 🫡

Since our last roundup we’ve added a new AI model, made Chat way better at knowing which words you’re talking about, and fixed a batch of bugs you reported. Here’s what’s new.

🆕 New model: Kimi K3

We’ve added Kimi K3 to Write, Draft, and Plugins. Kimi K3 is a new AI model praised for its creative writing chops, sitting at #2 on EQ-Bench’s creative writing leaderboard (second only to Claude Opus 5, which we added last week). It’s apparently quite good at register: ask for hardboiled or lush and it changes its sentence rhythm, rather than just adjectives. Audition it on a scene with a strong voice and let us know how the prose lands.

↩️ Better, safer Undo

We improved the way Undo handles Chat-based edits. If Chat had edited your manuscript earlier in a session, pressing Ctrl/Cmd + Z later could roll the document back to before that Chat edit instead of undoing only more recent manual edits. Undo now reverses the most recent change chronologically, the way it should.

💬 Chat reliability improvements

We’ve been very focused on improving how Chat handles your instructions.

• No more “your manuscript is empty” - It was possible for Chat to write you a useful chapter critique… and then discard it in favor of telling you your manuscript was empty. We fixed that!

• Highlights stick to the correct document - When you highlighted text and asked Chat about it, it was possible for Chat to reference a different file in your sidebar—or claim it couldn’t load your chapter at all. We fixed that too, and it should better understand that highlights are associated with your active document.

• Accurate word and mention counts - Ask how long a document or project is, or how many times a word appears in a document, and Chat now actually counts instead of estimating.

• Pasting editable text into Chat - Pasted text was turning into an uneditable chip instead of dropping into the message box as plain, editable text. We changed that so it remains editable.

• Long requests don’t vanish - A big Chat request that ran for many minutes could end in a bare “Load failed” with nothing saved, forcing you to start over. We made long-running requests like that preserve their work.

• Comment replies are reliable again - Sometimes @Sudowrite replies in document comments could fail with a generic “I wasn’t able to generate a reply” error. We’ve resolved that issue.

🛠️ More improvements & fixes

• Complete sentences from Brainstorm - Long Brainstorm ideas were being truncated, leaving sentences incomplete in some cases. You’ll get a complete idea from now on.

• Worldbuilding trait edits persist - We fixed an issue that caused Worldbuilding traits to overwrite themselves as you typed into them. (Autosave was a little overeager, but it’s tamed now.)

• Smart Dictation (on mobile) preserves nicknames - Inside a project, Smart Dictation was “helpfully” expanding a spoken first name or nickname into a character’s full Character Card name. It now writes the name as you said it.

Thanks as always—and please keep the bug reports coming.

Anthropic just released Claude Opus 5, their newest Opus-class model, and we've added it to Sudowrite.

According to Anthropic, Opus 5 comes close to the intelligence of their flagship Fable 5 model at half the price. It's a big jump over Opus 4.8 on reasoning and instruction-following benchmarks—which, for authors, means better Story Bible adherence and less plot drift. Early testers also report that Opus 5 may be more permissive than other models.

We haven't put Opus 5 through its paces on long-form fiction just yet, so we’re not ready to attest to its prose. The Opus line has historically been a favorite among Sudowrite authors for its warmth and character voice though, and on paper this is the strongest one yet—but you'll be finding out alongside us. Claude Opus 5 is available now in Write, Draft, and Plugins. Give it a spin and let us know what you think.

Since our last update we've focused on speed, reliability, and a few more model options. Here's what's new.

⚡️ A faster, more responsive Sudowrite

This week we made the Sudowrite web app load faster and feel snappier, especially on lower-end devices and slower connections. We optimized things so that both the homepage and the editor start up quicker, and you spend less time staring at a loading screen. Even scrolling now uses 60% less of your computer's processing power than it used to!

🆕 More model options: GPT-5.6 (Sol, Terra & Luna)

We've added OpenAI's newest GPT-5.6 model suite—Sol, Terra, and Luna—to Sudowrite, giving you even more AI model options so you can find the voice and style that fits your story. Try them today in Write, Draft, or your very own Plugins.

💬 Smoother, more reliable Chat & editing

We resolved a handful of issues you reported with Chat to help make it a better writing partner.

• Edits insert more reliably - We reworked the flow that applies AI rewrites into your document, so approved changes land where they should far more consistently.

• No more code or "thinking" text - Chat will no longer blurt out its internal reasoning or stray code in its replies.

• Streamlined conversations - Prose analysis and change descriptions now read naturally in the conversation instead of showing up as boxes in the chat. Chat will no longer derail a planning discussion just because you mention a chapter number.

✍️ Draft improvements

We fixed an issue where chapter prose could stop generating after the first scene (or in some cases, after 8+ scenes), and improved how Draft handles longer scene instructions so generations consistently complete as expected with all Prose Modes. If you do encounter an issue with Draft when using an experimental AI model (like the new GPT-5.6 suite), please submit a bug report so we can fix that.

🛠️ Even more improvements & fixes

• Story Bible now stays put - We tweaked Story Bible so that it stays open or closed based on your preference, rather than popping open on hover.

• Clearer sidebar icons - We’ve added more descriptive icons for the collapsible columns on either side of the editor, making it easier to restore your documents, history, or chat from a collapsed state. We also fixed a bug with resizing the editor when those are collapsed.

• Dark theme chat toggle visibility - We fixed an issue where the Chat Only / Allow Edits toggle in Chat was hard to read in dark themes.

• Selection menu clipping - We fixed a longstanding issue with the menu that appears when you highlight text, where if it was too long (or your screen was too small) it was clipped by the editor’s frame. It will now extend beyond the edge of the editor if necessary.

• Unpublish shared documents - We resolved a document sharing issue that prevented the ••• (more) menu from opening. That means you can once again unpublish a previously shared link.

• Trait text entry fix - We fixed an issue that was causing lag or text deletion while typing in Character and Worldbuilding trait fields.

We’ve got even more on the way. Keep the feedback coming!