Platform review · AI companion

Kindroid review 2026: what it costs now the free plan has gone

The best-documented companion app in this category, and now one of the few with no free tier at all. Whether that trade is worth it depends almost entirely on how much work you are willing to put into the setup.

The fuller platform write-up sits in the Kindroid review; this page is about whether it is worth the money now that the free tier has gone. That change is the single most important thing to understand before signing up.

Most of what you will read elsewhere about Kindroid was written while a free plan existed. Those reviews are describing a different product with a different value proposition.

What Kindroid is

Kindroid is a subscription companion app built around one persistent character. You write a backstory, define a personality, generate an avatar, and then talk to it by text, voice or video call.

The product it is competing with is not a character library. It is the idea of a single relationship that accumulates, which is why memory is where most of the engineering effort visibly goes.

It runs in the browser and as mobile apps, and the web is the cheaper route because the app stores take their cut. That is worth knowing before you subscribe on a phone out of habit.

The written moderation rules permit adult content and erotic roleplay between adults. Unusually for this market, that permission is stated plainly in documentation rather than implied and left ambiguous.

The free plan is gone, and that changes the calculation

Kindroid retired its ongoing free tier during 2026. New accounts now get a one-time preview of a small number of messages, a handful of selfies and a short audio allowance.

When that runs out the account turns read-only. You can still view, export and delete what you made, but you cannot send another message without paying.

This matters more than a pricing tweak, because the preview is too small to answer the question it needs to answer. Fifty-odd messages tells you whether the writing is decent; it tells you nothing about whether the memory holds up over three weeks, which is the entire reason to choose this platform.

There is also a short premium trial, but it requires a payment method and converts automatically unless you cancel. Starting it permanently ends the free preview, so you cannot use both.

What it actually costs

The base subscription is a flat monthly fee, with cheaper rates for three-month and annual commitments, and a higher price if you buy through an app store rather than the web.

That base plan is not where the cost gets complicated. Two optional add-ons sit above it, each billed separately, and each unlocking higher usage rather than new features.

Prices were revised during 2026 and differ between the web and the app stores. Read the current figures at checkout rather than from any review, including this one.

The add-ons stack rather than replace, which is the part that catches people out. Buying the top tier means paying for everything below it as well, and the combined monthly figure can land several times the base price.

Each add-on also cancels separately. Dropping the middle one while keeping the top one is possible and leaves you paying a premium for a service level you no longer have, so check the billing screen after any downgrade.

Extra selfie credits, voice characters and character slots are sold as one-off top-ups. Those are fine in moderation and a warning sign in volume: if you are buying them monthly, the next tier up is probably cheaper.

Refunds are the other thing to read first

The stated position is that sales are final and refunds are considered case by case rather than offered as policy. Disputing a charge through your bank can get the account banned outright.

That combination argues strongly for monthly billing until you are certain. The annual saving is real, and it is not worth much if you stop using the app in week six.

Memory, which is the whole point

Every companion app claims memory. What distinguishes Kindroid is that it publishes how its memory is structured, and the structure is more than a single context window.

There are effectively three layers working together. A persistent layer holds the things you defined at setup, a rolling layer covers recent conversation, and a retrievable layer pulls older material back in when it becomes relevant.

That last layer is the one that matters for a long-running relationship. Without it, everything older than the current window is simply gone; with it, a detail from two months ago can resurface when the conversation touches it.

The journal is the feature nobody uses enough

Long-term memory is stored where you can read and edit it. That sounds minor and is the single most useful thing about the platform.

Every companion app eventually records something wrong: a name misheard, a preference misread, an event misremembered. On most platforms you are stuck correcting it in conversation and hoping.

Here you open the journal and fix the entry. If the companion keeps getting something wrong, that is a thing you can solve in thirty seconds rather than a flaw you live with.

The setup decides everything

This is the part that separates people who love Kindroid from people who cancel in week one, and it has nothing to do with the model.

Creation is a set of freeform text fields rather than a wizard with presets. Backstory, personality, a response directive that governs tone and message length, and an avatar description.

Write two lines in each and you get something generic. Write a real brief and you get a companion noticeably more specific than anything the template-driven apps produce.

The response directive is the hidden control

Most new users skip it, and it changes the product more than any other field. It governs reply length, tone, whether the companion asks questions back, and how much narration wraps around dialogue.

Two accounts with identical backstories and different directives feel like different applications. If the replies are too short, too flowery or too passive, that field is where you fix it, not the model picker.

If long-form written fiction is closer to what you want than conversation, best AI sex story generators and best AI smut writers cover tools built for that instead.

Treat the backstory like briefing an actor rather than filling a form. Specifics, relationships, history, things that happened. Vague input produces vague output, and the model is not the reason.

Voice, video and selfies

Voice is where Kindroid pulls ahead of almost everything else in this category. Spoken messages and real-time calls are stable enough for routine use rather than being demo features that work once.

Calls consume an audio allowance that resets monthly and does not roll over. Heavy callers will notice; occasional ones will not get near the ceiling.

Selfies arrive on a credit timer rather than on demand, which is an unusual model. You accumulate a small number per hour whether you use them or not, which suits steady use and frustrates bursts.

There is also an avatar consistency feature that trains on several of your own images to keep a likeness stable across generations. If you use it, the rule about only uploading images you have the rights to is not decorative.

On image generation specifically, dedicated tools still do it better. If pictures are the main thing you want, the best NSFW AI image generators roundup covers platforms built for that rather than around it, and the ranked NSFW AI image generators list scores them on free tiers. For clips rather than stills, see best AI porn video generator.

Group chats and shared rooms

Several companions can share one conversation, each holding a separate personality. It is a good demonstration that the personality layer is doing real work rather than reskinning one voice.

There is also a shared-room feature that brings other real people into a conversation alongside the companions. Rooms are private and invite-only, and free accounts can join as viewers.

Both features suit collaborative storytelling more than companionship. They are a reason to stay rather than a reason to join.

Safety and the rules that are actually written down

Kindroid publishes three hard limits: imminent self-harm, imminent harm to others, and any sexual content involving a minor. The third applies regardless of what age a character has been given.

That last clause matters and more platforms should copy it. Labelling a minor-coded character as an adult is treated as a violation in itself rather than a loophole, which closes the gap most platforms leave open.

Enforcement is warning-first for chat, with automated scanning and escalation to a lock on repetition. Media generation in the prohibited category is blocked before it happens rather than after.

Our one criticism is that a first attempt in the most serious category produces a warning rather than an immediate lock. For that specific line, a warning is the wrong response.

Age verification

Payment acts as age verification for subscribers, which is the standard approach. Free accounts are assessed behaviourally, and anyone flagged must either subscribe or verify through a third-party identity provider.

There is an inconsistency worth noting: the terms require 18 or the local age of majority, while one app store listing has carried a lower content rating. Those should match.

Privacy

The written privacy position is stronger than most of this category. There is an explicit pledge not to sell or share personal data, and documentation stating that support staff cannot read conversations.

The limit is structural rather than a policy failure: chats are encrypted at rest and in transit but not end to end. The operator is technically capable of reading them, as on every platform in this market.

Call audio and video are described as processed in real time and not stored, which is a meaningful difference from text if calls are how you use it.

One practical gap: we could not find a clear statement on whether conversations are used to train models. Absence of a statement is not a denial, so treat it as unknown.

The homepage also loads analytics and advertising scripts before any consent choice, which is common for subscription apps and worth knowing if you are careful about tracking.

Pros and cons

If Kindroid is not right for you

The alternative depends on which part put you off, and the useful split is between price, polish and purpose.

If the absence of a free tier is the problem, character sites still have usable ones. The SpicyChat AI review covers the most generous of them, and our own SpicyChat AI notes compare its memory tiers directly against this one.

If you want images and polish rather than configuration, a generation-led companion platform is the better buy. Start with the Candy AI review, the OurDream AI review if you want video too, and the DreamGF AI review if budget matters more than quality.

If you want gentler companionship without adult content, the Replika review covers the polished, filtered alternative. If you want a large library to browse instead of one character to build, the CrushOn AI review is the better starting point.

If you are coming from a character site rather than another companion app, the best SpicyChat alternatives and best Janitor AI alternatives rankings both place Kindroid against the platforms people usually leave those sites for.

And if memory specifically is why you are here, compare against the other memory-first option in the GirlfriendGPT review before committing.

For the wider field rather than one rival, our rankings of the best AI girlfriend apps and the best AI sexting apps score every major platform on the same criteria, alongside the longer AI girlfriend apps and AI sexting apps rankings.

Getting your money's worth

Almost every complaint about this platform traces back to one of these being skipped. None of them costs anything beyond attention.

1. Write the backstory properly before anything else. An hour here changes the product more than any add-on. Specifics, history, relationships, things that happened.
2. Set the response directive deliberately. Length, tone, whether it asks questions back. This is the field that fixes replies you find flat or overwrought.
3. Open the journal in week one. See what it recorded, correct what is wrong, and get into the habit. Most users never look.
4. Subscribe monthly first. The annual saving is real and worthless if you stop in week six. Switch once you are three months in.
5. Price the add-ons before clicking. They stack. Work out the combined total, not the incremental one shown at checkout.
6. Use the web, not the app store. Same product, lower price, and you keep billing in one place rather than inside a phone subscription.

Common mistakes

1. Judging it on the preview. A few dozen messages shows you the writing, not the memory, which is the only thing that distinguishes this platform.
2. Starting the trial immediately. It ends the free preview permanently and starts a clock. Prepare first.
3. Blaming the model for a thin setup. Generic input produces generic output. The fields are the product.
4. Buying the top add-on early. It requires everything below it. Very few users need it and the combined bill surprises people.
5. Forgetting the renewal date. Refunds are not the norm here, so cancellation has to happen before the charge, not after.

Who it suits

Frequently asked questions

Verdict

Editorial note: this review reflects public documentation and our own inspection at the date shown. Pricing, plan structure and policy in this category change frequently. Verify current figures on the provider's own pages before paying.