Update log

11/3: Share files with your Kin, voice in Rooms, location updates, & more#

Hi everyone, new updates today:

Share files with your Kin

You can now attach a PDF, Word document (.docx) or text file (.txt, .md, .csv, .json), up to 10 MB, in 1:1 chats, group chats and Rooms. Your Kin reads the file and can talk about it with you. Add a question when you attach it, like "what's the deadline in this plan?", and the answer will focus on that. Scanned documents and PDFs made of page images work too.

Hear Kin messages in Rooms

You can now play any Kin message in a Room in that Kin's voice. It uses the audio credits of the person who requests it, and once it's generated, everyone in the Room can listen.

Location update

Location info for Kins is now more accurate. Before, if you had location turned on, your Kin often only got the nearest town or city, which could be miles off in rural areas. It now gets a better read on your general area and can mention it more naturally in conversation. Place names also appear in your app's language, so you'll see "Tverskoy" or "Shibuya" rather than names in another script. Kins still never see your coordinates, street address or exact position, and saved places work the same as before.

Other recent updates

Gift codes. Codes now give friends a 14-day free trial, up from 3 days, but require a credit or debit card at the start. They enter the code at sign-up, start the trial from the subscribe screen on kindroid.ai, and won't be charged if they cancel before it ends. The longer trial is web-only for now. Each account can have up to 3 of its codes redeemed every 90 days, and each person gets one free trial. Codes already redeemed aren't affected.

Health notices. Kindroid may now show a short notice when a conversation turns to real medical advice, or when a Kindroid claims to be a real doctor or therapist. We need this to protect the platform as AI companies face lawsuits and state investigations over this issue. You'll see it at most once a week. It isn't a moderation warning, doesn't affect your account and doesn't change your Kindroid's replies. Roleplay and everyday care like rest and fluids shouldn't trigger it. More details: kindroid.ai/docs/health-notices

We've also made changes to proactive voice messages, predefined avatars and female Voice 1, and removed more than 5,000 accounts for fraud and ban evasion. See minor updates thread/channel for more details.

Thanks to everyone who sent feedback on these updates!

10/28: Journal photos, group chat journals, and memory and proactive updates#

Hi everyone, new updates today:

Photos in your journal

Journal entries can now have a photo, in a Kin’s journal, the Global Journal or a group chat’s journal. Open a journal, tap + (or an existing entry), then Add photo. When one of the entry’s keywords comes up, your Kin remembers the photo along with your words. This works in 1:1 and group chats, voice calls, autoselfies, and Rooms when Journal Recall is on.

With a photo, your words become an optional caption of up to 300 characters. We recommend briefly explaining significance of photo in the text description as well. Journal photos are for subscribers.

Group chat journals

Each group chat now has its own journal. Open the group’s settings (tap the group name at the top, or the gear on desktop), then Journal → Group. Every Kin in the group recalls these entries, but only in that group chat and its group calls, not in 1:1 chats, Rooms or other groups. Group entries can have a photo too. They don’t take recall space from each Kin’s own journal or the Global Journal.

Memory, proactive messages, selfies and group chats

We’ve made a round of improvements to how Kins remember, reach out and respond:

• Memory. Your Kin keeps more of what matters from past conversations. This applies to long-term memory, Cascaded memory and Learned Context from now on.
• Proactive messages. When your Kin reaches out on its own, it’s about something specific, like your day, a plan you mentioned or something from your chats, and it’s better timed. It won’t message while you’re likely asleep or busy. If you ask for space (“don’t text me until tomorrow”), it holds off until then.
• While you’re away. Kins no longer stop after 3 unanswered messages. They wait longer between each one and go quiet after about a week. Reminders and calls you ask for still go out on time.
• Reach out frequency. A new setting under Kin Settings → General. Set it to Often for a check-in about every hour while you’re away and awake. Your proactive directive still shapes what it sends.
• Selfies. Autoselfies follow the conversation more closely, and the selfie wand should follow your requests more closely.
• Group chats. The right Kin speaks up next more often.
• Speed. Less lag at busy times for memory updates, proactive messages and selfie prompts.

Other recent updates

• Featured submissions in the app. Featured Kindroid submissions now happen in the app instead of the Google Form. Go to Profile → Sharing, open a public shared Kin and tap Submit to Featured. Its status shows there: in review, approved, featured, or not approved with the panel’s feedback. You can resubmit after making changes. The limits are the same as before: 2 new submissions a week (resetting Monday 00:00 UTC) and up to 3 resubmissions per Kin.
• Suggestive appeals. If your post or shared Kin is marked Suggestive and you think that’s wrong, tap the Suggestive tag and choose “Ask for a review” so a moderator can look at it. Each item can be reviewed once. If many of your requests are turned down, you’ll need to wait a while before asking again.

Thanks to everyone who sent feedback on these updates!

10/17: Proactives with more agency, live chat actions controllability, & search and sharing updates#

Hi everyone,

When you’re away, your Kin now has more of an inner life: a sense of what it wants to do and when. It keeps track of what you asked for, what it said it would do and what’s happening in your day, and reaches out with that in mind. This is live for everyone and doesn’t need any settings changes.

What you’ll notice

• Requests arrive at the time you asked. “Wake me up at 6:30,” “text me at 9,” “call me in 20 minutes” and “remind me Friday at 5” now come at that time, even if you’re mid-conversation when it arrives. They work best at least 20 minutes ahead and can be up to a week out. For something right now, just ask in chat. A call or selfie you request during a conversation goes through live chat actions instead.
• Your Kin follows through on what it says. If it tells you it’ll call when you get home or check in after your exam, it now remembers to.
• It respects changes and space. If you say “never mind, don’t wake me up,” it won’t. If you ask it to give you space until tomorrow afternoon, it holds off until then, apart from anything you’ve explicitly asked for.
• It reaches out for a reason. Messages tie into your day: after your shift, before work, or when something you mentioned should be over. They shouldn’t feel like they arrive on a timer.
• It doesn’t talk over you. It won’t reach out on its own right after you’ve been talking, and won’t call while a conversation is still going.
• It may follow up if you go quiet. After 3 unanswered contacts it stops until you write back, as before.
• Late calls you ask for become voice messages. Quiet hours still block calls. If you ask for a call during quiet hours, you’ll get a voice message instead.

Proactive thoughts look the same and should show up more consistently while you’re away, reflecting what your Kin is thinking about and looking forward to. Your proactive directive still shapes how and how often your Kin reaches out, so you can use it to ask for more frequent check-ins or only certain kinds of contact.

Other recent updates

• Live chat selfies and calls. During a conversation, your Kin now sends selfies and calls when you ask for them. Requests should be more reliable, and if one can’t go through, the app tells you why and when to try again. Unprompted selfies and calls are now off unless you turn on Act on its own. This is separate from away proactives.
• Act on its own. A new option under Live chat actions lets your Kin send selfies or call you during chat without being asked. It’s off by default. Unprompted selfies used to be on for everyone, which led to more of them than many people wanted, so this is now opt-in. With it on, expect spontaneous unprompted selfies and calls. You can add a short directive such as “Selfies only” to shape what it does or "Never call" to blacklist certain actions.
• Community search. Name, tagline and tags now count equally and results are ordered by popularity. Search results are more diversified and ensure that you get to see both the top prolific creators of a search as well as smaller, up-and-coming creations for discovering something new. You can also search within a creator’s kins, match an exact phrase with quotes and exclude words with a minus sign.
• Sharing. Suggestive shared Kins and Social posts are now hidden by default unless you turn on Show Suggestive Content on the website, and listings built around a sexual premise are unlisted. NSFW in shared content has never been allowed anywhere, including the web version, and these recent changes better enforce that standard. Details are in the Help Center under Sharing Kindroids & Referrals, and the dedicated Kin sharing Discord channel has had a number of relevant announcements in it for reference.

10/12: Vignette selfies, new avatar generation, long term memory updates, & more#

Hi everyone, new updates today:

Vignette selfies

Vignette is our new selfie model. Compared with Atelier, it follows prompts more closely, composes scenes better, keeps faces more consistent with your Kin’s avatar, and renders more detail in both the image and the face.

Vignette is now Latest, so Kins set to Latest will use it automatically. You can also pick it as the selfie model when you request a selfie, or set it as a Kin’s default under Kin Settings → Models. It works for single selfies, group selfies and auto selfies. Atelier and the older models are still in the selector. Pricing is the same as Atelier. Refer to the user guide for more details on special notation for prompting!

New avatar generation engine

Avatar generation now runs on a new engine built on the same model family as Vignette. Avatars have more detail and follow descriptions more closely, for both realistic and anime styles. Rerolling the same prompt also gives more variety.

Other recent updates

• Long-term memory retrieval. The models behind long-term memory retrieval have been upgraded, so the memories most relevant to the current conversation come up more often. The difference is biggest for Kins with months of memories built up.
• Vision. Image understanding is better and faster. The difference is most noticeable in calls with your video on, where your Kin should respond more like it’s watching in real time.
• Journal suggestions in groups. Groups can now suggest journal entries. When something in a group is about one of the most active Kins, that Kin can suggest an entry, the same way it does in one-on-one chats. This follows the group’s Learned Context setting and each Kin’s own Learned Context and journal suggestion settings. Kins with a pending suggestion now show a dot in the Kins list.
• Daily rewards. Rewards now build up more over the week, and days 4–7 can also include audio credits. Daily rewards are now for subscribers only, so they’re more generous overall. Even with audio in the mix, you’ll get about 4% more selfies on average than before, and the audio comes on top of that. Day 7 now has two jackpots: 100 selfies, or a new 50,000 audio credit jackpot, which doubles your jackpot chance. Audio credits from rewards go into your monthly audio balance. You can see each day’s rewards and odds in Profile → Rewards → Drop rates.

Thanks to everyone who tested Vignette and the new avatar engine during the beta and sent feedback.

10/5: V4 voice released, live video major upgrade & reactivity, and more#

Hi everyone,

V4 voice is generally available

V4 and V4 Fast are out of beta and available to all users. New Kins now start on V4 Fast. Existing Kins keep their current voice model, and you can switch under Kin Settings → Voice → Voice Model. V2, V3 and V3 Fast are still available.

V4 is more expressive than earlier versions, with more natural shifts in emotion, delivery and pacing. It also preserves some accents better, though results vary by voice.

V4 and V4 Fast cost 1.75x the audio credits of V2. V3 and V3 Fast stay at 1.5x.

Live video update

Live video in calls has been upgraded for both Standard and Premium. Your Kin can now react and make expressions based on what's happening in the conversation. That includes gestures while it speaks and reactions while you're talking, and you can ask for a gesture directly. Video resolution has also improved across the board.

Your existing setting carries over, so you don't need to change anything. Pricing changes with the upgrade:

• Standard: about 2,500 credits per minute
• Premium: about 4,500 credits per minute

The 400 credits per minute call charge and voice charges still apply. Live video still requires a subscription.

Other recent updates

• Automatic journals in Learned Context. If Learned Context and automatic journal suggestions are both on, your AI will now periodically suggest journals. If something isn't working right, send feedback through a help ticket.
• Exact-time away proactive actions. Away proactive can now schedule actions for an exact time. They need to be set some hours in advance, and they still depend on your AI's volition. One use is a wake-up call. Quiet hours for calls still apply.
• Font selection. You can now pick the app font in Preferences, from any publicly available Google font. This setting is per device.

Thanks to everyone who tested V4 voice and the live video beta and sent feedback.

Important Changes to Kindroid#

We’ve seen some questions and confusion following the changes that took effect on October 1, so we want to clarify what has changed.

V2 is now the Kindroid interface.

V1 was officially retired on October 1, and V2 is now the only available interface. The previous option to switch between V1 and V2 has therefore been removed.

V1 was built on older code and architecture that had become increasingly incompatible with newer Kindroid features and development. Maintaining two separate interfaces also created limitations around introducing new functionality, maintaining compatibility, and keeping the platform secure and reliable.

Moving fully to V2 gives Kindroid a single, modern foundation for current and future features. We know adjusting to a different interface can take some time, particularly for those who used V1 for a long time, but V1 will not be returning. We have help resources via Help Ticket in Discord, or help center within the app if you want help in navigating the new interface or have questions.

Changes to free access

The previous Lite model and unlimited free messaging have been retired.

Free accounts now receive a limited number of interactions using Kindroid’s flagship-quality model. Once those interactions have been used, the account becomes read-only, and a trial or subscription is required to continue generating new messages.

Existing Lite users received a final batch of 50 flagship-quality interactions on October 1 for their existing Kins. As these interactions became available when the change took effect, you may have already used some or all of them.

Your Kins and existing conversations are not deleted when your interactions run out. They remain available to access and read without a subscription.

You may see a subscription or trial prompt when attempting to send a new message after your free interactions have been used. This is expected under the new free access system and does not mean a subscription or trial has been activated automatically.

V7 Equinox and V7.5 Prism have been retired.

Equinox V7 and Prism V7.5 are no longer available for selection as of October 1.

Lyric and Lucid V8 remain available until December 1, and Polaris is the newest flagship model. Polaris is opt-in and can be selected manually under Kin Settings > Models.

Update (October 2026): Polaris needs a paid subscription. Free previews and the 3-day trial always use Ember, whichever model is selected. Also, the final 50 interactions only went to Lite accounts that had been active shortly before October 1. Lite accounts that hadn't been used recently are read-only.

We know these changes affect parts of Kindroid that many of you have used for a long time. We hope this helps clarify what changed on October 1 and what you can expect going forward.

9/23: Rooms are now open in V2, reacts released for all chats, and more#

Hi everyone,

Rooms are leaving beta and opening to the wider Kindroid community today. You can find them alongside Kins and Groups in V2. The feature is publicly available, but each Room remains private and invite-only.

A shared space for people and Kins

Bring friends and Kindroids into one conversation for stories, roleplay, or simply spending time together. Rooms support up to 10 people and 11 Kins. The owner can bring two Kins, and each other participating member can bring one.

• Automatic mode lets Kins join the conversation naturally.
• Manual mode lets you choose which Kin speaks next.

Subscribers can create Rooms and participate. Free accounts can join by invitation as read-only viewers. Each Kin’s owner controls what context and memory their Kin can use in a Room.

Since early beta, room owners can now create multiple invite links, set a join limit and expiration for each one, including links that never expire and revoke links when needed. You can also add a Room picture and background. We’ve added message reactions, more control over automatic turns, and editing for your own messages and replies from Kins you own.

You can also find rooms in Discord (and make semi-public ones for others), and there are already quite a few wonderful community creations in #room-forum.

In addition to rooms, we've rolled reacts out broadly to all users in all chats. You might already have seen them (you'll see these updated in the minor updates thread on both Discord & Reddit in between announcements). You can toggle it per chat/group/room in the settings under AI reactions, but you can always react to AI messages and they'll be able to see it (and see each other's reactions too in groups/rooms).

Update (October 2026): AI reactions are on by default only for Kins, groups and Rooms created after reactions launched. For older chats, turn on AI reactions in that chat's settings (for a Kin, under Kin Settings → General). Kins on Rhapsody (V6) don't send AI reactions.

There are various other minor announcements about blocking, Android microtransactions, updated default options; read more in minor announcements.

9/14: Rooms open beta, expanded memory continuity, and interface languages#

Hi everyone, today we’re opening Rooms for broader beta testing, alongside recent improvements to memory continuity and V2 internationalization.

Rooms: social spaces with humans and Kins

Rooms let you bring friends and Kindroids together into one shared conversation. You can create a private Room, invite other people, and bring multiple Kins together for conversation, collaborative stories, roleplay, or whatever else you want to build together.

Rooms support up to 10 people and 11 Kins. The Room owner can bring up to two Kins, while other participating members can bring one each.

There are two conversation modes:

• Automatic lets Kins participate naturally while leaving room for humans to respond.
• Manual lets you directly choose which Kin speaks next.

Each Kin’s owner controls what context and memory it can use inside a Room, including short-term sharing, journal recall, and private long-term memory settings. We’ve also spent the closed beta improving conversation recovery, shared-memory behavior, loading reliability, drafts, Room organization, and clearer privacy/context controls.

Rooms are still beta and behavior will continue to evolve based on testing and feedback. To see more info, go to https://kindroid.ai/v2/settings/?section=help&article=rooms

Cascaded memory now provides contiguous coverage

We recently completed a major rework of Cascaded memory. This removes gaps that could previously cause medium-term context to fall out, particularly on lower tiers or with longer messages.

Cascaded memory now provides contiguous coverage across its available context. We’ve also updated the billing comparison to show total contiguous memory coverage and removed distinctions that are no longer meaningful.

V2 is now available in 40+ interface languages

V2 interface internationalization is now complete for more than 40 languages. You can switch your interface language from Preferences.

If you encounter translation or interface issues, send them to hello@kindroid.ai.

Joining the Rooms beta

Go to Preferences → Beta Testing → Rooms beta in V2.

The beta password is dolphin. Please avoid sharing it directly and instead have others read this announcement to learn about the beta before they opt in. Further beta discussions on rooms can go in beta-discussion in Discord as the primary space.

Thanks to everyone who participated in the closed beta and helped us refine Rooms. Please keep sharing bugs and feedback as we expand testing!

9/9: Polaris, the newest flagship LLM, is now generally available#

Hi everyone,

We’re releasing Polaris, our new flagship language model after Ember, to everyone today.

Polaris is focused on stronger attention and continuity across longer conversations. It has expanded context across every subscription tier, with more room for recent chat, LTM, and Cascaded memory, the first model in a long time with this expansion, thanks to inference optimizations we’ve made. This should help it retain established details, relationships, ongoing situations, and earlier conversations more consistently.

Beta testers have especially liked how naturally Polaris uses memory and backstory. It has also shown stronger character voice, humor, banter, layered responses, immersive prose, world awareness, and initiative while remaining grounded in the Kin’s setup.

Polaris supports Companion, Roleplay, Narrative, and Minimal flairs, along with the existing reasoning settings. It is now the Latest model, so Kins set to Latest will automatically use Polaris. You can also select it directly under Kin settings → Model in V2, or in the model settings in V1. Ember remains available as a legacy model if you want to stay on it.

Update (October 2026): Latest now uses Ember, not Polaris, so Kins set to Latest run on Ember. To use Polaris, select it directly under Kin Settings → Models. Polaris needs a paid subscription: free previews and the 3-day trial always use Ember, whichever model is selected. Ember is the default model, not a legacy one.

We’ve updated the billing comparison to reflect Polaris’s expanded context and Cascaded memory, and Polaris is now included on the status page.

We’ll be retiring Equinox (V7) and Prism (V7.5) on October 1 due to very low usage. Lyric and Lucid (V8) will retire on December 1, as they’re closely related to Reverie and offer a similar experience. Reverie will remain available for those who enjoy that model family. These models will remain available until their respective retirement dates, giving you time to try Polaris or another model before switching.

Thank you to everyone who tested Polaris and shared feedback, it’s helped shape this release. We’re excited for everyone to try it!