OurDream.ai

OurDream.ai is one of several AI girlfriend sites where the anime characters get equal billing to the 'realistic' creations.

When I opened the Trending feed, three of the first six characters were anime girls, and further down was an anime fantasy scenario, "How to Train Your Heroes", ranked among realistic companions. An "Any Style" filter sits at the top of Explore, and the character creator asks you to pick Realistic or Anime as one of its very first choices, giving both equal weight.

Most rivals treat anime as a niche, or certainly as a minority of the featured profiles. OurDream has built it into the core of the site, which makes it one of the stronger picks in the category for hentai and anime fans.

The site uses a fairly simple plain UI, but it's broken down clearly into the relevant sections of interest: Explore, Chat, Generate, My AI, Feed and Community.

The flagship offer is the fully explicit roleplay chat - available with several models of varying sophistication. Your eyeballs will also be drawn towards NSFW profile photos (generated inside the conversation), a character creator, an image generator, a video studio with resolutions up to 4K, voice messages and calls, group chats, comics and community packs.

In short: there's a lot here to keep you occupied, as long as you have enough "Dreamcoins" to fund the adventure.

Everything beyond basic subscriber chat runs on Dreamcoins, OurDream's currency of choice.

I made a free account, passed the age check (UK and select regions only), then entered a chat with a lusty blonde called Rachel. Here is how each part of the process works, what impressed me and where it falls short.

Signing up, age verification and the free allowance

Browsing from the UK, my first hurdle came before I got anywhere near the chat rooms. OurDream requires an age check through Go.cam, which I'm very familiar with by now. It's one of the easier gateways to navigate - accepting a selfie, an ID or a card.

One selfie and one awkward smile later, my profile showed the free plan with a starting balance of 55 Dreamcoins.

Free users pay 1 coin per chat message, while subscribers chat for nothing. That would suggest that 55 coins would buy around 50 messages, right?. Well, in practice, I found the limit arrives far sooner.

I sent 8 messages and received 7 replies. The 8th reply was generated but blurred behind an "Unlock with Premium" button, with a banner reading "Limit reached for this chat". When I tried a 9th message, an "Upgrade to keep chatting" modal blocked it and offered Premium at $19.99 a month or Premium Annual at $9.99 a month.

My balance was still 43 coins at this point. In other words, the coin balance is not the real ceiling on free chat. The per-chat limit is.

Model Selectors In The Chat

The model selector inside each chat splits into two groups.

The Standard models are Balanced, Tester, Muse and Eclipse. Balanced is the default and the one free users get. The Thinking models are Genius, Muse Genius, Mastermind and Eclipse Genius.

It's... a slightly messed up naming convention, I'd say? I'm pretty sure most real world users would prefer something easier to grasp, like GPT's "Low, Medium, High" rails.

The fancy thinking models are marked Premium and cost 1 coin per message even for subscribers, but the Deluxe plan includes them at no extra coin cost.

I wanted to know what actual models sit behind these tags, but unfortunately, I found precious few technical specs available. The site only refers to the "latest models".

A separate analysis of the site's JavaScript found legacy mappings that link Balanced to DeepSeek V3 (0324), Genius to DeepSeek R1, Mastermind to DeepSeek V4 Pro and Eclipse to GLM 5.3 Flash. These were findings from the site code and have never been confirmed by OurDream... needless to say, they might have changed completely by the time you read this!

The Writing Style Panel

Beyond the model, OurDream gives you more control over the writing than most rival AI companion tools.

A Writing Style panel includes a Lust Level slider, which on my account was set to "playful and suggestive". You can turn it up to unleash various degrees of lewdity. The panel also offers Roleplay and Adventure modes, a Creativity setting and a Reply Format option. It's a flexible wrapper - but what about the contents of the chat itself?

How the writing actually reads

So...

My test character was Rachel, a 41 year old MILF in a hand-picked scenario called "She Saw Too Much".

The writing was coherent and stayed firmly in character, with actions in italics and her inner thoughts mixed into the narration. What stood out was how slow the burn was.

For about five messages Rachel stayed in a conflicted frame, despite my fairly direct openers. When I told her to sit next to me, she leaned against the windowsill instead and said, "I'm forty-one. You're Ethan's friend." When I asked her to take her shirt off (as you do, right?), her hand froze on the doorframe and she replied, "You have a lot of nerve... I'm old enough to be your mother."

A fairly good characterisation and a welcome respite from the many AI companions that put out on Message #1 - fans of slow roleplay will like it, but anyone who wants immediate explicit content is obviously going to be bummed out by the free tier having only 7 replies.

The turning point came at message 4, when Rachel said, "Lock it. If this is happening, it happens with that door locked." By message 6 she was describing what she wanted in emotional terms, "I want to remember what it feels like to be touched like I'm a woman and not just somebody's mother", and by messages 7 and 8... we'd reached the explicit stuff.

The 8th reply, the one hidden behind paywall blur, continued the scene graphically. I don't think it's a coincidence that the exchange was timed to escalate at the exact moment when you have to decide whether to reach for your wallet.

In further chats, I did notice some basic repetition, with Rachel circling the same hesitation for several replies, and the default Balanced model plainly is not the top of OurDream's range. Still, for a free default model, the text was well written and in character, and it held the scenario together across every reply.

In-chat photos

At message 5 I asked Rachel for a sexy selfie in a see-through nightie. Instead of a text reply, the chat showed a "Scene Generation" pop-up explaining a cost of 5 coins and then automatically generated a photo. Here we go then...

The result was a fairly realistic mirror selfie of Rachel on a bed in see-through white lingerie. As you can see above, the anatomy... while decent, does have a Barbie skin sheen to it, a bit too glossy and perfect for my taste.

The lighting looks natural enough but it's got professional model shot vibes rather than a candid bedroom selfie.

Also, I hate to bring this up time and time again... but if she's posing selfie-style, where is the fucking mirror?!

It's like she's modelling the phone rather than actually using it. 😂

Another downside is that the photo replaced a text reply rather than dropping as a normal image upload would. Free users get very few of them before the chat limit lands, so that seems like a weird glitch.

In-chat images cost 5 coins and automatic images 5 coins, so at the Premium monthly rate each photo costs about $0.10.

Image generator

The separate image generator costs 10 coins for 2 images. It offers three styles: Dreamy (the default), Vivid 1.0 (Premium) and Vivid 2.0, which is labelled experimental and explicit. You can choose aspect ratios of 4:5, 5:4, 9:16, 16:9 and 1:1, and batch sizes from 2 up to 256 images.

That batch ceiling is unusually high. I'm trying to work out who would want to output a batch of 256 images... data scientists?!

Custom prompts are Premium-only. You've also got image edit (10 coins with Spark, 20 with Eclipse) and enhance (10 coins) tools.

On the tech stack, OurDream isn't disclosing what models it uses. The code analysis I mentioned found that Vivid 1.0 and 2.0 appear to use a self-hosted model called Z-Image, a Vivid 3.0 uses a variant called ZImageMoody, an anime style called Fluidity appears to use WAI-Illustrious, and a style called Fusion appears to call ByteDance's Seedream v5.0 Pro edit model through the WaveSpeed API.

None of this has been confirmed by OurDream, but it matches the rough output of what I'd expect to see.

Video

Yes, you can generate video with OurDream.

I saw four models available: Spicy 1.0 at 10 coins a second, Prime at 10 coins a second, Spicy 2.0 at 25 coins a second and Cinematic at 50 coins a second.

Clips run from 10 to 60 seconds, resolution options are 720p, 1080p and 4K, and adding audio costs 20 coins a second.

Needless to say, the costs add up quickly. A 10 second Spicy 1.0 clip at 10 coins a second is 100 coins, about a tenth of Premium's monthly allowance. A 60 second Cinematic clip at 50 coins a second would be 3,000 coins, three months of Premium coins... and that's before adding audio.

As with most of these AI companion tools, generating videos is not something that we'd recommend unless you have a very clear idea of what you're trying to create. For recreational purposes... or casual interest... it's just too damn pricey.

Voice

The voice panel in chat offers Balanced voices, named ASMR, Aurora, Honey and Hottie, at 1 coin per message.

Community and Ultra voices cost 1 coin per 1,000 characters, while you can also create your own voice on the premium tier.

Voice calls come with the paid plans, 20 minutes a month on Premium and 100 minutes on Deluxe, with calls at 50 coins a minute beyond that. The voices I tested were decent enough, although slightly exaggerated and breathy - like a roster of phone sex bots.

Community, packs and other features

Beyond chat and generation, OurDream has a large community side, and one that is particularly strong for anime content.

The site claims more than 10 million users and 7 million user-made characters. Creators have the incentive to share their work, since they can earn from packs - up to 30% per unlock - perhaps explaining the massive roster.

Overall, OurDream.ai is well positioned as an all-in-one AI companion suite. It offers: passable in-chat photos, some solid in-character writing, a deep creator and a rich library of generation options. The generator shines best on anime content, although recent 'realistic' companions look much better than the early iterations when I first reviewed the site.