Changelog

DeepSeek's newest Flash is in Private mode, free on every plan: it counts toward Free's 30 messages a day, is unlimited on Pro and above, and plays characters on Free too.

E2EE mode now offers Qwen3.8 27B and Qwen3.6 35B, run by NEAR AI Cloud. Your browser checks NEAR's secure enclaves itself and verifies a signed receipt on every reply.

Two models from Meta, in Anonymous mode.

Google's Gemini 4 model, with low, medium, and high effort.

GPT-6 Astra now offers all five effort levels.

Gemini 3.8 Flash in chat, and the Nano Banana image models in Studio, on Pro and above.

Choose how much GLM 5.3 and the uncensored models think before they answer.

Sort your own characters into folders on Custom. Search for a creator to move all of theirs at once, or open a character to move just that one. Search, sort, and folders now show as soon as you have one character, not only from 12.

Character chats show their privacy mode at the top, as other chats do. Click it to open the chat's settings.

The Export for Morse bookmark now works on Chai, JanitorAI, and Chub as well as Character.AI. Click it there to save your chats, the characters in them, and your personas, then import the file in Morse.

Featured's For you shows the characters you've met, the Halloween guests in a row of their own, and the people they can introduce you to next. Halloween guests in the spotlight wear a pumpkin. Every world is still a tap away at the top.

A chat you start without an account holds a shorter conversation than a free account's. When it gets too long, you can sign up free right there, and the chat comes with you.

The import page has a tab for each place you can move from: Character.AI, Chai, JanitorAI, Chub, and other apps' files. Each shows its own steps, and what comes over from there and what doesn't.

On Morse's front page, the message box, its suggestions, the bar that follows you down the page, and Start free are now frosted glass over the sky.

The welcome page is shorter, and now shows the nine worlds you can role-play in, each playing its scene as you open it and introducing a character who lives there, and what you can make in Studio.

Plus is now called Pro, and Pro is now called Max. The plan that was called Max, which is no longer sold, is now called Ultra. The plans table says more plainly what each plan includes.

Free includes 15 messages a day, and its character chats have no daily limit. Memory, end-to-end encrypted chats, uncensored models, and deep research now come with Pro, which also includes Nano Banana 2 Lite and Qwen Image 3 Pro.

The API docs moved to docs.trymorse.ai, with search and a playground that sends a request from the page with your key. Old trymorse.ai/docs links take you to the same page there.

When you try Morse without an account, character chats now show how many free messages you have left, and when they run out you can sign up for free and keep going right where you left off.

Halloween's characters wear a pumpkin in the corner of their card, and their page says plainly that Pro members and up keep them after Halloween ends on Nov 7.

The plans on the welcome page and in Choose a plan are now one table, with each feature on its own row and a check or a cross under every plan. On a phone, pick a plan above the table to see what it includes.

The welcome page has a new look: a dawn sky, in light mode or dark. It shows what each privacy mode sends to a model's provider, how an end-to-end encrypted chat reads on your device and everywhere else, and why your chat history stays on your device. You can switch it between light and dark at the bottom of the page.

The page for bringing your Character.AI characters over now opens on the same morning sky as the welcome page.

The top of every chat now shows how private it is, and after your first reply it says what your message was sent without. The privacy tour waits until you've had a couple of answers, and its highlight lines up with the model menu.

The blog's articles open on the same morning skies as the welcome page, in place of the dotted landmarks, and end with the welcome page's footer.

Once you have a dozen characters of your own, Custom has a search, and you can sort them by Recent, A to Z, or Most chats. Long lists load as you scroll, so a big import opens quickly.

Create a key in Settings, under API keys, to call Morse from your own code with any OpenAI-compatible client. Requests made with a key are paid from your credits. The developer docs at docs.trymorse.ai show how to send your first request.

The exporter on the import page is open source. You can read its code on GitHub and check that it's the same file Morse gives you.

When your browser hasn't agreed to keep Morse's storage, the sidebar says so and asks it to, so it won't clear your chats when space runs low. If it still says no, the sidebar tells you what to do in your browser, or you can be reminded in a week.

If you came to Morse from one of our ads on X, we now tell X whether you signed up or paid, naming you only by the ID X put on the ad's link. The privacy policy explains what's sent, and turning on Global Privacy Control in your browser opts you out.

Settings, under API keys, lists your requests from the last 30 days: the model, the key that sent it or the Morse app, the tokens, and what it cost in credits. Copy a request's ID from its row to quote it to support.

Community opens on Trending, the characters people chatted with and liked most this week, and has a row of genres to browse. A character's page says more about them and lists their genres, and tapping who made them shows the rest of their characters. When a character you've chatted with gets an update, you can switch your chats to the new version.

Community opens as a gallery, each character's picture large and sharp, as Featured's are. Switch to List for more of each in words, and Community remembers which you picked.

Sharing a character now starts with its genre picked when its name or description says it, like a vampire or a mafia boss, for you to keep or change. Community lists its genres most popular first, and hides the empty ones.

Your characters in Community are credited to your own handle, like @riley, that no one else can use, and you pick it the first time you share. Tap a creator's handle to see all their characters, with their messages and likes.

Tap the heart on a Community character to like them. Cards show how many likes each character has, Popular puts the ones with the most messages and likes first, and Liked lists the ones you liked. You can also show only male or only female characters.

Community cards and pages show how many messages people have sent each character, and Popular and Trending rank characters by them.

Our team looks over every character shared to Community before others can see it, usually within 24 hours. Until then it says In review, and if it isn't approved you'll see why and can edit it to send it back.

A character from Community, or one of your own, now has the NSFW switch, privacy, and who you play as right on its page before you begin, as Featured characters do. Its description and first messages call you by the name of whoever you play as.

A new character's last step has Share in Community. Turn it on and saving takes you straight to sharing, where you see what goes public before it does.

New character opens on a form to write your character yourself: a picture, a name, their notes, and how they open the chat. Write it with AI is still there when you'd rather describe them in a few lines.

Featured's All view has a filter at the top, so you can see only the male characters or only the female ones. It remembers your pick.

Import takes character cards (PNG or JSON) and files from SillyTavern, CAI Tools, and other apps, not only Character.AI, and points you to our guide for saving your characters and chats from Character.AI.

Morse has a new logo: a four-pointed star split down the middle, in navy and bright blue. It's in your browser tab, on your home screen, and beside the name, and it lights up while a reply is on its way.

The Character.AI import page, the changelog, and the legal pages now share the home page's bar, star, and serif headings.

Share any character in your library, including ones you imported from Character.AI. They stay credited to whoever made them, and one that's already in Community opens there instead of being shared twice.

Community has a Share a character button that lists the characters you made, and every character has Share.

Once a chat has read a web page, saving or forgetting a memory, or editing a photo you attached, waits for you to press Allow on it, which says what it will do, such as the memory it forgets. Writing your next message no longer lets it through, since the page could have written those words for you.

Our Terms now cover the Morse API: keeping your keys safe, paying with credits, and building for others. The app asks you to agree to them the next time you open it.

The E2EE verification report opens with what it proves in plain words, and marks each claim by whether your browser checked it itself or the gateway vouched for it.

Bring your Character.AI characters, every chat with them, and your personas to Morse, and pick up where you left off. Our exporter saves everything in one run, on a computer or a phone, and goes on from where it stopped if it's cut short; then import the file from Characters, or from Settings under Your data.

Make a character of your own under Characters, Custom: describe them, add a picture, and chat with them on every plan.

The model picker lists the models on your plan first, then the rest in one list, with small icons that show which use credits or need a plan. On a phone it rises from the bottom of the screen, with rows that are easier to tap.

Pick Anonymous, Private, or E2EE from tabs at the top of the model menu, next to the model you choose, and the list shows that mode's models. The privacy switch is gone from the top of the page.

The privacy tour's last card sits in the middle of the page, and Start chatting puts the cursor in the message box.

Memories that come over from your other device wait in Settings, Memory until you accept them, and chats don't use them before then. A transfer also leaves this device's Memory switch as it is.

Share a character you made with everyone, and find characters other people shared under Characters, Community. Your chats with them stay private, and if someone stops sharing a character or changes it, the chats you already started carry on as they were.

Moving your Character.AI chats from a phone shows each tap on an iPhone in a picture, and Android has steps of its own.

Back up your chats to an encrypted file from Your data in Settings, and restore it on any device you sign in to. Or send your chats straight to another device of yours, end-to-end encrypted.

Read about private AI and how Morse works on the new blog.

Branch a conversation into a new chat. When you edit a message or retry a reply, the earlier versions are kept.

After its first reply, a chat gets a short title that sums up what it's about.

A chat that sends end-to-end encrypted stays in E2EE mode, so a later message can't go out less private by mistake.

Ask chat to change an image it made, or a photo you attached, and it edits that image instead of starting over.

In E2EE mode, GLM 5.3 Flash runs on NEAR AI Cloud first. When NEAR AI is down, your message goes to Phala instead, still end-to-end encrypted and verified on your device.

Press Ctrl+K (⌘K on a Mac) to jump to any chat, project, or setting, and Ctrl+/ (⌘/) to see every shortcut. Tooltips show each button's keys.

Morse can remember short facts about you, kept on your device, and use them in later chats. It saves lasting facts as you chat, asks first before keeping anything sensitive, and lists them all in Settings under Memory.

Moving from ChatGPT or Claude? Import what they remember about you from the Memory section in Settings.

A Search button at the top of the sidebar opens chat search, so you can find a chat on your phone too.

Chat from the home page without an account, and attach files to your first message. Your chat stays in the sidebar, and signing up keeps it and lets you start more.

Every plan can now search the web and read pages, in any chat whose privacy mode allows it.

Open What's new from your account menu to see what's changed in Morse, month by month. A dot shows when there's something you haven't seen.

Wherever Morse lists the privacy modes, it uses the same words and goes from least to most private.

Install Morse as an app from your browser. Morse also asks your browser to keep your chats stored on your device.

Every chat gets the tools its privacy mode allows, with no separate agent mode to turn on.

E2EE mode checks more carefully that your messages only reach a genuine secure enclave.

The model menu marks models you don't have enough credits for. If a message can't be sent, Morse tells you why.

Updated type, corners, and depth across the app.

After setup, a short tour explains the privacy modes.

Featured rows have arrows and scroll with your mouse wheel or touchpad.

Meet Hesper, a new character.

Invite friends to Morse and see how your invites are going. Your invite link takes a friend straight to sign-up, they can also enter your code by hand, and the plans dialog tells them when an invite is waiting.

Copy a message you sent, not just the replies.

A redesigned phone layout, with chips above a one-line composer.

A Safety page, and a crisis support card that appears on your device when you may need it.

Drag a chat into a project, and preview the files you've sent.

Meet characters through letters of introduction. Their letters arrive in a mailbox on Characters.

Send us feedback from Settings.

The model picker is one searchable list grouped by maker. Hover or tap Credits to see what a model costs.

See your purchase history and receipts in Settings. A credit top-up is confirmed as soon as you come back from checkout.

A Morse link you share shows a proper preview, and an invite link shows your invite card.

Buy credits straight from the plans dialog.

When a new version of Morse is out, the app offers to reload.

When our Terms change, the app shows you what changed.

A refused image now says the image model refused the prompt, and suggests rephrasing it. On Free, a model that uses credits says so when your balance is empty, and offers to switch that chat to the model Free includes. If your email isn't verified, the notice offers to sign out, so you can sign in again with a code or Google.

Once a reply has searched your other chats, checked the weather, or looked up a crypto price, saving or forgetting a memory and editing an attached image wait for your Allow, as they already did after reading the web. Text in those results could have come from someone else, so the model can no longer act on it by itself.

Imported characters keep their tagline as their tagline instead of having it written into their character notes, including Character.AI characters with no description.

Moving the pointer onto the model you've already chosen no longer slides a highlight in from the row you left.

The encryption report's limit on a Phala model server's code now says this browser didn't check what software the server booted, and that the compose and OS image it shows are the server's own report. Before, it said the browser saw which software the server runs.

The privacy badge at the top of a chat no longer lights up and swaps its name for a sentence after your first reply. It just shows your privacy mode.

A reply that reached this browser from a backup or another device keeps its verification report. The lock under it stays closed, with a small history mark, and says the reply was encrypted when the device that received it checked it. This browser can't re-check a reply from a backup or another device, so the lock no longer claims it did, and the report says the checks are that device's record.

Chats saved with a current SillyTavern now import under their character's name, joined to the character's card, rather than as a character called "unused". Group chats are left out.

When you switch a chat from Private to Anonymous before it has a title, the model that names it now reads only what the Anonymous reply itself was allowed to read, so an earlier answer that used your memories or other chats stays out of it.

A file of characters with one entry Morse can't read imports the rest, rather than none of them.

Importing a character card with the same name as one of your characters now adds it as a new character, unless it's the same one. Before, it quietly went into the character you already had, so it seemed to disappear.

If you have thousands of characters, the sidebar now lists them a page at a time and stays quick, and a big Character.AI import no longer slows down as it goes.

On a phone, the privacy status at the top of a chat no longer pushes the top bar past the edge of the screen after your first reply. Character chats leave it out, since their privacy is set in Chat settings.

Moving the pointer onto the model you're using no longer slides its blue highlight in from another row. It stays on your model the whole time.

Opening a Community character you've chatted with lists your chats, with Continue and New chat, instead of starting over. You can also pick which of its first messages a chat opens on.

Deleting your last chat with a character from Community deletes them from your device too. A character you edit or delete changes in Community right away, and Share waits until a character has a first message.

On a computer, the phone steps for moving from Character.AI show the iPhone's, and pressing Stop during an import no longer skips your personas. A dropped folder that can't be read now says so.

A portrait you painted for a new character stays when you then write the character with AI, instead of being painted, and paid for, again. Starting a chat the moment a character's page opens keeps your NSFW setting.

Opening a creator's page from one of their characters shows all their characters, even after you searched, picked Male or Female, or were looking at your Liked ones.

A character from Community you chat with no longer shows up under My characters. It's in the sidebar with your chats, and its page says what its sharer sees: how many chats it has, never yours.

Opening Morse in a new tab no longer restarts the loading star partway through. It plays on until the app appears.

Opening Morse in a new tab no longer flashes the sidebar over the loading star before the app appears.

Morse asks before it erases the chats on your device. If it's told your account was deleted, it keeps them until you choose to erase them.

After you delete your account, the chats on this device are erased even when another tab still has them open. The erase finishes once that tab lets go, or the next time Morse opens, and until then Morse says so instead of waiting.

If your new account couldn't take over your guest session when you signed up, your guest chats are kept and move over the next time you sign in.

With Morse added to your iPhone's home screen, the top of the screen no longer shows a blurred smudge of the buttons under it.

Words you type on the welcome page before it finishes loading are kept, and you can send them.

When every model fits in the model list, the last one no longer shows faded out, as if there were more to scroll to. This showed on iPhones with the page zoomed out.

The model menu on a phone no longer shows a sideways scrollbar under its list.

When you send your chats to another device, you check that the numbers match only on the device sending them, and an animation shows your chats on their way.

Deep research reports no longer show up as raw code.

When sign-in is busy, Morse waits and tries again by itself instead of saying it can't reach the server. On a phone turned on its side, the model menu opens beside its button.

The app no longer gets stuck loading when a message can't be decrypted.

Web search no longer fails in Private mode.

Ask a model which model it is, and it answers correctly.

Signing in on a new device no longer starts the setup again.

If Google sign-in fails, you land back in the app instead of on an error page.