FunFun AI Review: Strong Presets, a Picture Button and a Paid Year

On paper it does three jobs: companion chat, image generation and video generation. I'd rank them in reverse order of how loudly the site sells them. The pictures come first, the video is the upsell on the upsell, and the conversation is the frame around both.

The iOS app sits at 1.0 out of 5 from 2 ratings. Two ratings prove very little, and nobody should lean on them. My own verdict is kinder than that score and a good deal less kind than the sales copy.

Is it worth paying for? For a narrow reader, yes: an adult who wants a pretty preset companion, likes asking for photos and doesn't mind paying before seeing much. For anyone who wants long, careful roleplay, a free tier you can actually judge the app by, or a price list before checkout, I'd spend the money somewhere else.

Key takeaways: - The picture button is the product, and the chat mostly exists to lead you to it. - The free tier is a short demo that ends at a paywall within a handful of messages. - Prices are hard to find, and the yearly discount is designed to look like the only sane choice. - Credits run out long before the month does if you like pictures, and video costs far more than stills. - Read the refund and renewal rules before paying, because both are tight and both favour the house.

The yearly plan in the table says more about the pricing page than about my judgement. It would be nice to call it a careful value decision. It was not, and the pricing section explains how the page is built to produce exactly that result.

The table shows the mismatch plainly. That gap between active weeks and a paid year is the most useful thing I can pass on. A yearly plan on an app you use for a few weeks is a gym membership in January.

A note on what is current. The facts about today's FunFun AI were checked on 23 September 2026 on the web version with a free account. That check covers the sign-up, the free tier, the settings, the paywall and the public legal pages, not a paid account.

So the split is simple. How the chats, the pictures and the credits felt comes from October 2025. What the plans cost, what the terms say and how the sign-up works now comes from that September 2026 check, and each section says which is which.

That was the moment the whole business model made sense to me. The free tier blurs images, and that one blurred frame is the entire sales strategy in a single picture. I'd call it honest in a way, because it didn't pretend to be anything else.

What greets you today

The site is live at funfun.ai. Its page title reads "FunFun AI: AI Girlfriend | Chat & Get Pics". I like a title that tells you the deal up front, and this one does: chat, and get pictures.

The interface is available in 16 languages. That's more than many bigger apps offer, and I think it's one of the few places where FunFun AI is more generous than it needs to be.

The sign-in page and the side door

The front door is oddly shut. The sign-in page shows nothing but a Cloudflare Turnstile box asking you to "Verify you are human". On 23 September 2026 that check did not pass on its own, so the page led nowhere.

There is a side door that works. Typing a message into a character's chat as a guest opens a Sign in / Sign up window, and that window needs no captcha. I'd find it funny if it weren't so typical: the proper entrance is locked, and the one next to the product is wide open.

The window offers two ways in, email or Google. Email sign-in is passwordless, so the site sends a magic link instead, with the message "Check Your Email - A sign in link has been sent". There is no password at any point of sign-up.

Clicking the link logs you in and drops you back into the same character's chat. It is a neat piece of flow design. You were in the middle of saying something, and the app returns you to the sentence.

What it doesn't do is ask your age. Sign-up has no age or birthday question, and no age confirmation appears at email sign-up. For an app with a mature side, I'd expect at least a checkbox at the door, and there isn't one here.

There is also no onboarding after sign-up. New accounts get an auto-generated handle made of two random words and a short code. It reads like a wedding table name picked by a random generator, and nothing prompts you to change it.

The login session persists between visits. So once you're in, you stay in, which is convenient for you and for anyone else who picks up your laptop.

The iOS app

FunFun AI also has an iOS app, listed as "FunFun AI: AI Chat & Roleplay". The seller on the App Store is CHAINDESK. That matches the company behind the website, which is reassuring in a low-key way.

The iOS app was released on 10 July 2025. It was last updated on 25 March 2026. Six months without an update isn't a scandal, but I read it as a sign that the web version is where the attention goes.

My overall read on the entrance: the fastest part of FunFun AI is getting in, and the slowest part is finding out what anything costs. That pattern holds for the rest of the review.

The layout is simple enough. The logged-in home page has the tabs Featured, Characters, Images, Videos and Leaderboard. The left menu holds Home, Chat, Gallery, Generate Image, Create New Character, My Characters, Premium, Discord and Contact.

The count isn't settled, though. A reviewer at callyourgirlfriend.com counted 337 characters, all photoreal, with no anime and no search. That is a long way from about 20, and I'd guess the difference is between the preset stars and everything users have added over time.

The missing search matters more than the number. With a few dozen faces you can scroll; with a few hundred you need a filter, and FunFun AI gives you tabs instead. Finding a specific kind of character becomes a flea market visit: fun for an hour, useless when you want one particular thing.

My captain on the leaderboard

The part I did use was the community side. There was a gallery of characters with a leaderboard of the popular ones, and once my own character existed I got hooked on it. I checked my captain's rank every morning like a stock price, coffee in one hand and no dignity in the other.

He peaked at a very modest position. I'd love to report a surprise rally, but a no-nonsense ferry captain was never going to top a chart built around photo requests. He had the market appeal of a well-kept ledger.

I still think the leaderboard is the best idea in the app. It gives creators a reason to come back, and it shows new users what everyone else is actually picking. It just can't make up for a shelf this narrow.

My verdict on the library: fine for choosing a pretty face in a minute, thin for anyone who likes to browse, and missing the search box that would make the rest of it usable.

The wizard asks for name, ethnicity, age, eyes, hair style and hair colour, among others. It is closer to a casting form than a character sheet. You describe the person you'd cast, and the app fills in the rest.

There is a gate in front of it. Creating a character is premium-only, according to the upsell that reads "Upgrade To Create AI Girlfriend". So the wizard is part of the pitch, not part of the free tier.

My Estonian ferry captain

My creation was a no-nonsense Estonian ferry captain. It is not the obvious choice for a companion app, and that was the point. I wanted to see whether the wizard could handle someone who wasn't built from the default presets.

He came out surprisingly coherent. He sounded like a man who checks the weather twice and the passengers once, which is exactly how I'd pictured him. After a lot of apps that turn every custom character into the same cheerful stranger, that felt like a small win.

I'd give the wizard credit for staying out of the way. It didn't try to upsell me on every screen, and it didn't bury the useful fields under decorative ones.

The Extra Context field

The most useful tool is the least visible one. Each character has an "Extra Context" field of up to 500 characters to steer the bot, with a small counter that starts at "0/500". The text is saved per character.

Five hundred characters is about a short paragraph. It's enough for a job, a temper and a couple of habits, and not enough for a backstory. I'd write it like a note to a second shooter: who this is, what they never do, and what the scene is about.

My view overall: the wizard is the second-best part of FunFun AI after the picture button. It's quick, it's clear and it produced a character with a spine. It just lives behind the paywall, and the bill for it arrives late.

She didn't melt. She roasted my hoodie for fifteen minutes without softening once. Every time I defended it, she found a new flaw in the fabric, the fit or my life choices.

Very professional, as I wrote in my notes. I have worked with stylists at shoots who were gentler, and a few who were not. Mio would have fit right in with the second group.

That consistency is the best thing about the chat. A tag on FunFun AI is a promise the character keeps, which is more than I can say for plenty of apps where every personality drifts back to the same polite middle.

How the replies read today

The preset "Lily Chen" is a dentist and an anime fan. Replies take about 5-10 seconds. That's slow enough to notice and fast enough not to matter, like a render preview that shows up just as you reach for your coffee.

Replies are written in roleplay style, with stage directions. Asked about her day, the dentist describes a root canal and then turns the question back on the user: "(smiling warmly, adjusting my scrubs) Oh, it was quite busy! Had to perform a root canal today... Did you remember to floss today?".

Asked what anime she watches, she names a comedy series she says she's rewatching. It's a quick, confident answer rather than a vague "oh, lots of things", which is the right instinct for a character built around a hobby.

My bot-to-bot argument

The moment I'll remember from FunFun AI had nothing to do with romance. I asked Mio the stylist to rate a photo of an outfit my ferry captain was wearing. Given what she had said about my hoodie, I was not expecting kindness.

Then I took her verdict to the captain and had him respond in his own chat. From that point I was the courier, and each bot only knew what I pasted in.

Carrying an argument between two bots by copy-paste felt exactly like relaying notes between a bride and her mother on a wedding edit. Each side was sure of itself, neither could hear the other, and I was the only one doing any actual work. At least the bots never asked for another revision round.

It says something that the best scene in the app needed me as a courier. FunFun AI has no shared room where two characters meet, so any group scene is a manual job. I'd still recommend trying it, because it shows how firmly each preset holds its line.

My overall read on the chat

The conversation is better than the rest of the product suggests. The presets are written with a clear voice, they hold their tags, and the roleplay style gives each reply a bit of scene.

What it lacks is depth. The replies are good lines rather than a story, and nothing in the chat pushes a plot forward on its own. For short scenes with a strong character, it works; for a long campaign, I'd look elsewhere.

An outside review reached the same result. A reviewer on intimeros.com found that short-term memory worked, with a stated preference recalled later in the chat. Two independent checks landing on the same answer is enough for me to trust the short-range memory.

The promises go further. The iOS description says "Your AI companions remember what you love, what you don't". The site copy says characters "evolve and adapt to your interactions".

I'd read both lines as marketing until proven otherwise. Remembering a root canal from five messages ago is continuity inside a scene. Remembering what you love across weeks is continuity across a whole project, and that's a different job.

In editing terms, the app keeps the props consistent within a scene. Whether it keeps them consistent between shooting days is not something any source shows, and there is no visible memory panel, journal or notes page to check.

The closest thing to a memory tool is the Extra Context field on your own characters, covered in the creation section. It's a fixed note, not a memory, but it's the one place where you control what the bot keeps in mind.

My advice is to treat FunFun AI's memory as good for a single evening and unproven beyond it. If a detail matters to you, put it in the Extra Context note rather than trusting the chat to hold it.

This is not the app I'd pick for a long relationship story. For short scenes with a strong preset, the memory is enough, and that's roughly the use the rest of the app is built for anyway.

On a free account, the Generate button stays disabled. The web Premium benefit list includes AI image generation. So the art tool is a shop window on the free tier, and you can look at the tools without using them.

Video: my captain's wave

The web Premium benefit list includes AI video generation as well. The AI Art Generator has an "Image to Video" tool. Third-party reviews report video generation too.

Video generation ate far more credits than images. I made one six-second clip of my ferry captain waving from the bridge, and it cost about as much as a week of pictures. That's an expensive wave for a man who never smiles.

The motion was mushy enough that I'd have rejected it as B-roll. If a camera operator had handed me that shot, I'd have asked for another take and a steadier hand.

What outside reviewers report

Outside figures describe smaller clips than mine. callyourgirlfriend.com reports 3- to 5-second video clips at up to 720p for 20 credits. My clip in October 2025 ran six seconds, so either the length changed or the reports describe a different setting.

aigirlfriendscout.com describes the feature as image-to-video animation rather than full scenes. That matches my own impression: a still that has been nudged into motion, not a scene with a camera move. I'd treat it as a moving photo and price it that way in your head.

Voice

Voice is the thin part. The Terms mention AI-generated voice messages, but no live voice-call feature appears on the official pages. A reviewer on intimeros.com saw no voice features at all and describes the experience as text-first.

For me that's not a loss, but it's worth knowing before paying. If you want calls, FunFun AI is the wrong app, and the voice options it does have are hard to find.

When the messages run out, an upgrade popup titled "Upgrade To Unlock Unlimited Messages" opens. A picture request can bring it on even sooner. Sending a picture request as the 4th action opened the same paywall.

The free account got no free image. That's the key detail. The picture button is the reason most people come to FunFun AI, and a free account can press it without ever receiving anything unblurred.

Blur and credits

All character gallery images are blurred for free users, each with an "Unblur" button. A free account apparently has 0 credits. So there is nothing to spend, and nothing in the gallery to see clearly.

I don't mind a strict free tier on principle. I do mind a free tier that looks like it's offering a sample and then hands you an empty plate. Here the plate is blurred, which is at least honest about what's on the menu.

What the upgrade popup promises

The popup's list is short and clear. It offers "Unlimited text messages", "Create your own AI Girlfriend" and "Remove image blur".

It also promises "Get 100 FREE credits / month", "Faster AI response time" and "Access to premium features". The last one is the vending-machine line of the list: it tells you nothing and still sounds like a reason to pay.

The word FREE in capitals, attached to credits you only get by paying, made me smile. It's the same trick as a free dessert with a set menu. The dessert is fine, but it isn't free.

I'd use those messages on purpose. Pick the preset you're most curious about, ask one question about her day and one follow-up, and watch the reply time and the tone. Don't waste one on "hi".

My verdict on free

As a way to use the app, the free tier doesn't exist. As a way to decide whether to pay, it's a very thin sample. I'd call it the weakest free offer I can think of among apps that still bother to have one, and I'd plan to pay from the first minute or not bother at all.

The monthly price was there to make you panic-buy the year. Nobody is meant to pay $25 for a month; they are meant to see $25, then see $5.99, and feel clever. It worked on me, which is the least flattering sentence in this review.

I panic-bought the year. The subscription spreadsheet I keep for exactly this reason is disappointed in me. If it could talk, it would say the page was simply good at its job.

The crypto option

You can pay with crypto. I didn't, but the option made me think about who the target audience is. A companion app that offers crypto is telling you who it expects to meet at checkout: people who would rather the purchase not appear anywhere familiar.

Today: prices behind a checkbox

On 23 September 2026, the web prices are hidden. The billing and pricing page shows only a Cloudflare Turnstile checkbox, so no prices are visible in the app. The /pricing and /premium pages returned HTTP 404 that day, and there is no public pricing page.

The billing page's own template offers monthly, quarterly and yearly subscriptions, with a line that reads "Save {discount}% with annual billing". That placeholder says a lot. The discount is so central to the pitch that it has its own variable.

The Terms offer a monthly or an annual subscription. A subscription includes unlimited messaging. That matches the upgrade popup and is the one plain promise in the whole pricing story.

What the outside world says it costs

Without a public price page, the only web figures come from other sites. A third-party review on pippinclub.com, unverified, lists $25 for the monthly plan. The same review lists $16.25 a month on the 3-month plan.

It also lists $5.99 a month billed annually, described as "70% off the $19.99 monthly rate". That is a different anchor from the $25 on the page in October 2025, which makes me think the discount story gets rewritten as needed.

A deal page on intimeros.com, undated, lists $25 monthly, $48.75 quarterly ($16.25 a month) and $72 yearly ($5.99 a month, shown as 70% off). That matches the October 2025 prices in my notes, so $72 is the best available figure for what my year cost.

Other sites quote lower starting points. On aigfbot.com, one Reddit user cites $12.99 a month for Premium, unverified. aigirlfriendscout.com lists paid plans from $12.99 a month and "Discrete billing". flirtygf.com lists a $5.99 starting price.

So the comparison is uneven. In October 2025 I paid $5.99 a month billed yearly on the web; the same yearly plan on iOS costs $89.99 today, or about $7.50 a month. The monthly plan went from $25 on the web then to $19.99 on iOS now.

How you can pay

The Terms list payment by card, Pay by Bank, PayPal, local methods or cryptocurrency. The web billing page shows "Instant Bank Payment" and "Pay with Crypto" options. flirtygf.com lists credit cards and crypto as payment methods.

That's a wide set of options, and I'd count it as a plus for anyone outside the usual card markets. The catch is that refunds work only for card payments, as the terms section explains.

Is the step up worth it?

If you'll use the app for longer than a couple of months, the yearly plan is the only price that makes sense. If you're unsure, I'd pay for one month even at the anchor price, because a month is a cheap lesson and a year is an expensive one.

My own year is the proof. The discount was real, the use was three weeks, and the maths only works if you ignore the calendar.

That makes the basic sum easy. At today's rate, 100 credits buy about fifty pictures a month and nothing else. It sounds like a lot until you see how easy the app makes asking.

Where my credits went

The "Ask for a Picture" button was right in the chat, so the flow was smooth. That smoothness was also why my monthly 100 credits were gone by day nine. I didn't plan it; the button just kept being there.

I'd compare it to a hotel minibar. Nobody decides to spend a small fortune on peanuts, but the peanuts are right next to the bed and there's no price tag in sight. By day nine my balance looked like the bill of a guest who checks out with regret.

It taught me the real price of the app: the subscription buys the chat, and the credits decide how much of the picture side you actually get.

Buying more

Extra credit packs are sold on a "Buy Credits" page. On iOS, 100 credits cost $9.99. 350 credits cost $34.99, and 550 credits cost $49.99.

The bigger packs keep going. 1150 credits cost $99.99, 2500 credits cost $199.99 and 3800 credits cost $299.99. Doing the math out loud: the smallest pack works out to about ten cents a credit, so a picture costs about twenty cents and a video about two dollars.

The biggest pack brings a credit under eight cents. I'd skip the top tiers entirely. Spending $299.99 on credits for an app with a thin library and a short video clip is like buying a lifetime pass to a diner you've only eaten at once.

Use them or lose them

Tokens must be used within their billing period, per the Terms: they "must be used within their respective billing periods". That turns the monthly allowance into a monthly deadline. Credits you don't spend simply disappear.

Combine that with a picture button that sits in every chat, and the design is clear. The app wants you to spend early, run out and buy a pack.

And if the monthly allowance is gone in the first week or so, as mine was, take that as information rather than a reason to buy a pack. It usually means the app has found your weak spot.

A new account also gets no free credits. The only email after sign-up is the sign-in link, with no welcome message and no coupon. That's oddly refreshing in a category full of countdown timers, even if it also means nothing to claim.

The yearly discount is the only deal

In practice the only discount is the yearly price itself. The "70% off" badge is a real saving against the monthly anchor, but it's a saving you pay for in advance, for a whole year. I'd call it a discount on commitment, not on the product.

No coupon code will beat it, because there's nowhere to enter one. If a site promises a special FunFun AI code, I'd assume it points to the same yearly plan with a nicer headline.

The affiliate program

FunFun AI runs an affiliate program on Rewardful at funfun-ai.getrewardful.com. It pays 40% of all referred sales for the lifetime of the customer, monthly, in its own words a "monthly lifetime Affiliate program to earn 40% of all the sales you refer". Affiliates also get their own affiliate Discord.

Forty percent for life is a generous cut. It explains a lot of the enthusiastic FunFun AI coverage you'll find online, and I'd read any glowing review with that number in mind.

My take on deals

There is nothing to hunt for here. The best price is the yearly plan, the second-best price is a single month, and every other offer points back to one of those two. I'd spend the time you'd have spent looking for a coupon on reading the refund rules instead, because those matter more to your wallet.

It felt like a bar where the band keeps asking if anyone wants to dance. You can stay at your table, but nobody pretends that's what the room is for.

How the switch is set today

Today the default is more careful. Mature content is switched off by default on a new account. Profile Settings has a "Show mature content (I'm over 18)" checkbox, and it is off by default.

For free users the setting hardly matters. Images stay blurred for free users whatever the content setting, and removing the blur is a premium perk. So the switch is really a paid-account control.

The sales copy still leans in. The homepage advertises "enticing selfies". It's a mild phrase, and I'd call it a fair summary of what the picture button is for.

What the Terms say about who may use it

The Terms require users to be at least 18 or of legal age in their country, in the words "users must be at least 18 years old or of legal age in their country". The Terms also claim that the site has an age gate.

The App Store rates the iOS app 17+. The Terms describe the service as one where "The APP includes AI-generated adult content for registered users only.". Those two labels sit a little awkwardly next to each other, and I'd trust the Terms over the store badge.

The filter

The blocked-content policy bans illegal content, hate speech and similar material. The moderation filter is based on the operator's own model, described as "a content moderation filter, based on our proprietary LLM technology".

The Terms also say this filter stops generated content from resembling minors. That's the right rule to have, and I'd want it stated just as plainly in the app itself.

A label that needs fixing

One detail in the character creator doesn't match that rule. The age options are a youngest option, then 20s, 30s, 40s, 50s and 60s. The youngest option carries a label that clashes with the platform's own age rule, even with a number beside it.

I'd rename that option before touching anything else on this page. It's a small string in a menu, but labels set expectations, and this one sets the wrong ones.

How strong is it, really?

Outside reviewers describe a softer app than the one in my October 2025 notes. callyourgirlfriend.com rates the content level as mild, not fully graphic. That fits the default switch today better than it fits my October 2025 impression.

My overall view: the settings are simple, the default is sensible and the blur makes free accounts a moot point. The age check is a checkbox and the creator's age menu needs a fix. I'd rate the content controls as adequate on paper and careless in one visible spot.

The Terms are governed by French law. They were last revised on 05/09/2024. That's two years without a revision, which is a long time for an app that has since added video and an iOS version.

Refunds

Refunds must be requested within 24 hours of payment. A refund is refused if the user has "used more than 20 tokens". Only subscriptions paid by card can be refunded.

Payments made by UPI in Indian rupees are not refundable. Credit packs are refused a refund once any of their tokens have been used. Put together, the refund window is a single day and a few pictures wide.

My advice is blunt. If you're unsure, decide within the first day and before you use the picture button much. After that, assume the money is gone.

Chargebacks

A chargeback leads to immediate cancellation. That's standard, but it's worth knowing that disputing a charge ends the account on the spot, together with everything in it.

Who owns what

The platform and its AI characters are owned by funfun.ai. Users receive a licence for personal use only. So your custom character is yours to chat with, not yours to take elsewhere.

Users grant FunFun AI a licence over what they type and upload, which the Terms call their Input. That licence ends when the Input or the account is deleted, except for content already used in commercial or sponsored posts.

That exception is the clause I'd read twice. Anything that has already made it into a promotional post stays in use after you leave. I'd keep personal photos and real details out of the chat for that reason alone.

Privacy

The Privacy Policy is dated 26 Apr 2024. It is very short, about 4k characters. That's roughly the length of this section, for a document meant to cover everything the app does with your data.

It says "We do not sell, trade, or rent your personal information". It also says data is shared with service providers. Both are normal lines, and the second one is where the detail would normally live.

Training and data use

The Privacy Policy says nothing about using chats to train AI. Whether chats are used for training is unclear; the policy says data is used to "Improve, personalize, and expand our website".

I'd read that as a door left open. "Improve and expand" can cover training without ever saying the word, and a short policy has room for that kind of vagueness.

My read on the terms

The company is real, the law is French and the rules are short. The refund window is narrow, the licence exception for promotional posts is wider than I'd like, and the privacy policy says less than it should. I'd call the terms tolerable for light use and a reason for caution with anything personal.

It reminds me of a client portal that says "Log in" while showing your files. Nothing is broken, yet you start doubting everything else on the page.

The bigger annoyance is the Gallery. The Gallery page keeps loading indefinitely. For a product that sells itself on pictures, a gallery that never finishes loading is an odd look.

I'd rank the frictions in order of how much they cost you. The blocked sign-in page costs you time, the endless Gallery costs you the best browsing view, and the "Login" header costs you a moment of doubt.

The pricing opacity belongs on this list too, even though it's covered elsewhere. An app that hides its prices behind a checkbox makes every other small bug feel less accidental.

My verdict: none of these would stop a determined user. Together, they make FunFun AI feel like a site maintained in bursts, with the checkout getting the most attention and the rest getting whatever time is left.

It also suits people who want crypto or PayPal payment, or a non-English interface. The payment list is wide and the language menu is generous, which counts for a lot outside the usual markets.

Who should skip it

It's not for anyone who wants a meaningful free tier, since free means 5 messages in total. You can't judge the app on that, so you're buying on faith.

It's not for anyone who wants a large community character library. The shelf is narrow, and without search it feels even narrower.

It's not for anyone who wants deep customisation or lorebooks. The wizard and a short context note are the whole toolkit, and they suit a sketch better than a saga.

It's not for anyone who wants transparent public pricing. If you like to read a price list before signing up, this app will frustrate you from the first minute.

My short version

Pay for FunFun AI if you want pictures of a preset companion and you're happy to learn the cost as you go. Skip it if you want to try before you buy, build a big world or know exactly what you'll be charged. Most people I'd point to one of the apps in the comparison table instead.

My captain is still waiting for passengers. Somewhere on a server there is a no-nonsense Estonian ferry captain standing on his bridge, checking a timetable nobody will use. I feel slightly guilty about him, which says more about me than about the app.

How renewal and cancelling work

Subscriptions renew automatically at the end of each period, in the Terms' words "Your subscription is set to renew automatically at the end of each subscription period". That applies to the yearly plan too, so an annual plan you forget about becomes a second year you didn't choose.

Cancelling is done in Settings with the "Unsubscribe" button. After cancelling, access lasts until the end of the paid period. So there's no reason to wait: you can cancel on day one and still use everything you've paid for.

Leftover tokens expire at the end of the period and do not carry over. A cancelled plan leaves no balance behind, only a date.

What happens to your data

After a subscription ends, conversation history and images are permanently deleted unless it's renewed. The Terms say "all your data (conversation history, images, etc.) will be permanently deleted after that date unless you renew your subscription before then.".

That means my captain's last voyage is already on the schedule. When the plan lapses next month, the chats and pictures go with it, and I'm at peace with that.

What to do before paying

Put the renewal date in your calendar the moment you pay. Better still, press Unsubscribe straight after paying, since access runs to the end of the period anyway. Save anything you want to keep, such as your Extra Context notes, before the plan ends.

Pick one month over the year unless you already know you'll stay. I'd rather pay the anchor price once than repeat an accidental year. If you do buy the year, press Unsubscribe the same day and let the access run out on its own.

Once you're in

Budget your credits like a rendering queue. Decide in advance how many pictures a month you want, and treat a video as a big job that needs a reason. The picture button will always be there, and that's the problem.

Use the Extra Context note on your own characters. It's the only reliable way to make a detail stick, and a tight paragraph there does more than any amount of reminding in the chat.

Keep personal photos and real details out of the chat. The licence terms and the short privacy policy don't earn that kind of trust, in my view.

Where to get help

The community lives on Discord, at discord.gg/8tRSbNz7Wn. Support is at [email protected]. For anything commercial, the business contact is [email protected].

If you run a site, the affiliate sign-up page is at funfun-ai.getrewardful.com/signup. The operator also runs a sister site, "AI Tools Directory", at funfun.tools.

No X, Instagram, Reddit or app-store links appear on the site. So don't expect to find the company on the usual social accounts; email and Discord are the channels.

Conversation quality gets a good mark. The presets hold their tags, the replies have stage directions and some personality, and the bots don't drift back to a bland default.

Memory is mixed. Short-range recall works within a chat, but there's nothing to show it lasts, and the only tool you control is a short context note.

The free tier is weak. Five messages and a blurred gallery aren't enough to judge the product, only enough to be curious about it.

Value for money is weak too. The yearly price looks cheap, but credits go fast, prices are hidden and refunds close within a day.

Images and voice land in the middle. The picture flow is smooth, the video is a moving still and voice barely exists.

Privacy and terms are mixed, with a real company and French law on one side and a thin policy and a wide promotional licence on the other. Cancelling is simple, but renewals and deletion rules need care. Content settings are mixed, with a sensible default and one label that needs fixing.

Where it does worse is almost everything around the money. The free tier is shorter than most, the prices are harder to find, and the refund window is about as narrow as they come. The library is small and has no search, which matters as soon as you want something specific.

My rule of thumb: if the table shows an app with a real free tier and a public price list, try that one first. Come back to FunFun AI only if you want quick pictures of a preset with a strong personality and you're fine paying before you know much.

In editing terms, FunFun AI is a stock footage site with good thumbnails and a confusing checkout. The clips are usable, but I'd never buy the annual licence on the first visit again.

What stays with me isn't the pictures. It's a stylist who wouldn't stop mocking a hoodie, and a ferry captain who argued about his uniform through a courier who happened to be me. The characters had more spine than the business around them.

So my captain waits on his bridge a little longer, and then the plan ends and he goes with it. I'll miss him slightly, which is a lot of feeling for a man made from a seven-step form.