HomeProductAutopilotAdsMCPWebsitePricingGuides
Sign inStart free
The AI inside Soclo

The best model for
each piece of content.

Soclo makes your content with Seedance 2.5, Wan 3.0 Prime, MiniMax H3 Max, GPT Image 2.5 and more — 18 models in all. You do not pick one, learn one, or pay for them separately. You write one line, and Soclo sends the job to whichever model is best at it.

Start free See the models

Checked against the engine on 17 September 2026. When a model changes, this page changes.

No accounts to set upNo separate billsOne line, whichever model it needs
Why it matters

One model is never
best at everything.

One holds a face across a whole clip. One films thirty seconds without a cut. One draws words you can actually read. One makes your own product look like itself. Choosing between them is a job, and it is Soclo's job, not yours.

  1. 1

    You write one line

    “Film a fifteen second advert for my new iced latte.” No model names, no settings.

  2. 2

    Soclo reads the brief

    How long it runs, whether a person is in it, whether words have to sit on screen, whether it starts from a photo you already have.

  3. 3

    It picks the model

    And tells you what it is making before it makes it. If a model refuses something, Soclo moves the job rather than quietly making something weaker.

Video

Which models does Soclo use to make video?

8 cameras from four makers, each better than the others at one thing. Soclo reads the brief and sends it to the one that suits it.

Swipe sideways to compare →

What each video model Soclo uses is best at, how long it films in one go, and whether it makes its own sound
ModelMakerLongest single clipPicture sizeMakes its own sound
Seedance 2.5ByteDance30 seconds720pYes
Wan 3.0 PrimeAlibaba30 seconds1080pNo — Soclo adds it afterwards
MiniMax H3 MaxMiniMax15 seconds768pYes
Kling v3Kuaishou15 seconds—Yes
Kling v3 ProKuaishou15 seconds—Yes
Veo 3.1Google DeepMind8 seconds1080pYes
Veo 3.1 FastGoogle DeepMind8 seconds1080pYes
Veo 3.1 LiteGoogle DeepMind8 seconds720p or 1080pYes

Seedance 2.5

ByteDance

Filming up to thirty seconds in one unbroken take, with sound.

Soclo uses it for
Longer adverts where a person has to stay the same person the whole way through. There are no cuts in a single take, so a face cannot change halfway.
Soclo picks another when
A short, sharp clip does not need thirty seconds, so Soclo picks a faster camera and saves you the wait.
ByteDance’s own page for Seedance 2.5

Wan 3.0 Prime

Alibaba

Clean 1080p motion, with a first frame, a last frame or reference photos to steer it.

Soclo uses it for
Adverts built from something you already have: a product photo, a shot of your shop, a picture of you.
Soclo picks another when
The clip needs its own soundtrack from the camera itself, which Wan leaves to Soclo to add at the end.

MiniMax H3 Max

MiniMax

Words on screen, and a lot of direction in one prompt.

Soclo uses it for
Offers, prices and short punchy lines that have to be readable as the clip moves, plus complicated briefs other cameras trim.
Soclo picks another when
The words have to be exactly right down to the last letter, in which case Soclo lays them over the film rather than asking a camera to draw them.
MiniMax’s own page for MiniMax H3 Max

Kling v3

Kuaishou

Holding one person’s face across a whole clip from a single photo.

Soclo uses it for
Putting you, or an actor you have permission to use, in the advert and keeping them recognisably themselves.
Soclo picks another when
The advert runs longer than Kling films in one go.

Kling v3 Pro

Kuaishou

The same identity work as Kling v3, with more detail in the picture.

Soclo uses it for
The hero shot of a person-led advert, where the face is the whole point.
Soclo picks another when
The shot is scenery or product, where the extra detail changes nothing you would notice.

Veo 3.1

Google DeepMind

Believable scenes built around a reference photo, with sound the camera generates itself.

Soclo uses it for
Building a whole scene around a person or a product that has to look like it was filmed somewhere real.
Soclo picks another when
The advert needs one long take, which Veo does not do in a single pass.

Veo 3.1 Fast

Google DeepMind

Most of Veo 3.1, back much sooner.

Soclo uses it for
Everyday posts, and trying an idea before committing to the slower version of it.
Soclo picks another when
The shot is the one you are paying to run as an advert, where the wait is worth it.

Veo 3.1 Lite

Google DeepMind

Quick, sound-carrying clips at a modest size.

Soclo uses it for
The routine clip in a week of posts, where fast and good beats slow and perfect.
Soclo picks another when
The clip is the centrepiece, not the filler.
Pictures

Which models does Soclo use to make pictures?

Drawing a product, drawing words and drawing a photograph are three different jobs, and no single model is best at all three.

GPT Image 2.5

OpenAI

Getting a real product to look like itself, and following an instruction exactly.

Soclo uses it for
Product pictures, and any picture where the thing in it is something you actually sell.
Soclo picks another when
The picture is mostly type, or the subject is one the model declines, in which case Soclo moves it to another model rather than quietly making something weaker.
OpenAI’s own page for GPT Image 2.5

Seedream 4

ByteDance

Text inside a picture, at very high resolution, quickly.

Soclo uses it for
Posters, price cards, menus and anything where words have to sit inside the image.
Soclo picks another when
The shot is a photograph of a person, where a photoreal model is the better choice.

Nano Banana 2

Google

Looking like a photograph rather than an illustration.

Soclo uses it for
Lifestyle shots, people, food and anything that should read as taken rather than made.
Soclo picks another when
Your own product has to be reproduced faithfully, which is a different job.

Nano Banana Pro

Google

The same photoreal look with more detail held in it.

Soclo uses it for
The one picture in a campaign that gets blown up big.
Soclo picks another when
The picture is a thumbnail nobody will ever see at full size.

Gemini 2.5 Flash Image

Google

Changing a photo you already have, from a plain-English instruction.

Soclo uses it for
“Make the background white”, “take the van out”, “make it evening” — on your own photo.
Soclo picks another when
Only one part of the picture should change and the rest must not move at all.

FLUX.1 Kontext [pro]

Black Forest Labs

Changing one thing in a photo and leaving everything around it alone.

Soclo uses it for
Swapping a label, fixing a sign, tidying one corner, without redrawing the shot.
Soclo picks another when
The whole picture is meant to change.

Bria RMBG 2.0

Bria

Cutting the background away cleanly, on a licence built for commercial use.

Soclo uses it for
Product cut-outs for a shop page, an advert or a catalogue.
Soclo picks another when
The background is part of the story.

Clarity Upscaler

Clarity

Making a small or soft photo big and sharp.

Soclo uses it for
Rescuing the only photo you have of something, so it can go on a poster.
Soclo picks another when
The picture is already large enough.
Voice and music

What about voice and music?

A voice that sounds like a person, and music written for the clip rather than borrowed.

ElevenLabs Turbo v2.5

ElevenLabs

A voice that sounds like a person reading, not a machine announcing.

Soclo uses it for
Voiceover and narration over a clip, in the voice you choose.
Soclo picks another when
The camera has already generated its own sound and a second voice would fight it.
ElevenLabs’s own page for ElevenLabs Turbo v2.5

Stable Audio

Stability AI

Music written for the clip rather than borrowed from a library.

Soclo uses it for
A soundtrack under a post, without a licence to worry about.
Soclo picks another when
The clip is somebody talking and music would get in the way.
Editing

And then you change it
yourself.

No model gets it right every time, which is why Soclo has a real editor rather than a "try again" button. Change a word, a shot, an image or a whole page, and keep everything you liked about the rest.

See the editor
Questions

The AI inside Soclo, answered plainly.

Which AI models does Soclo use?

Soclo makes video with Seedance 2.5, Wan 3.0 Prime, MiniMax H3 Max, Kling v3, Kling v3 Pro, Veo 3.1, Veo 3.1 Fast, Veo 3.1 Lite; pictures with GPT Image 2.5, Seedream 4, Nano Banana 2, Nano Banana Pro, Gemini 2.5 Flash Image, FLUX.1 Kontext [pro], Bria RMBG 2.0, Clarity Upscaler; and voice and music with ElevenLabs Turbo v2.5 and Stable Audio. Soclo picks the right one for each piece of content, so you never have to.

Do I have to choose a model?

No. You say what you want in one line and Soclo chooses. If you have a preference you can steer it, but the default is that the choice is made for you and explained in plain English.

Do I need my own accounts with any of these?

No. There is nothing to sign up for, nothing to configure and no separate bill. They are inside Soclo, on your Soclo plan.

Why not just use one model for everything?

Because they are good at different things. One holds a face across a whole clip, another films thirty seconds without a cut, another draws readable words inside a picture, and another reproduces a real product faithfully. Picking one for everything means being worse at most things.

Which model films the longest clip?

Seedance 2.5, which films up to thirty seconds in a single unbroken take. That matters because a face cannot survive a cut: every shot is a separate generation, so a person can come back looking like somebody else. One take has no cuts.

Which model is best for words on screen?

MiniMax H3 Max renders type well while the clip moves, and Seedream 4 is the strongest at words inside a still picture. When the wording has to be exactly right, Soclo lays the text over the finished film rather than asking a camera to draw it.

Which model is best for my own product?

GPT Image 2.5, which is the most faithful at making a real product look like itself and at following an instruction exactly.

Can I put myself in the advert?

Yes. Kling v3, Veo 3.1, Seedance 2.5, Wan 3.0 Prime and MiniMax H3 Max can all carry a person from a reference photo. Real-person likeness is consent-led and safety-gated, and a synthetic character is never presented as a real customer.

What does Soclo use to plan and write?

Soclo AI. It reads your Brand Kit, plans the week and writes in your voice. The models named on this page are the ones that make the content itself.

Will this list change?

Yes, and that is the point of it being Soclo’s job rather than yours. When a better model arrives it goes in behind the same one line you already type.

One line, the right model

You write the sentence.
Soclo picks the camera.

Free to start, on the App Store, Google Play and the web.

Start free →Ask from your own AI
Download on theApp StoreGet it onGoogle Play