MiniMax H3 Max
Five seconds of video in about three seconds, at the lowest rate in the studio.
Learn more
dGEN Visual StudioMCP & AI AgentSign inEvery top model for image
and video. One AI studio.
The all-in-one platform to generate and edit high-resolution images and videos with top-tier AI models. Maintain brand consistency using custom LoRAs, co-create with an intelligent AI assistant, or construct reusable node-based visual workflows for creators and autonomous AI agents via MCP.
Nano Banana Pro
Veo 3.1
GPT Image 2
Kling 3.0
Seedance 2.5Seedream 5
Flux 2 ProMiniMax H3
AI Image GeneratorCreate or edit high-end visuals from prompts, images or your own LoRA.AI Video GeneratorTurn your idea into cinematic video, with sound where the model can do it.
AI AssistantSay what you need and it writes the prompts. Name a model and it uses that one, or let it choose.
dGEN FlowCreate AI workflows on a visual canvas and reuse them again for every product, client or campaign.
Tools & TemplatesPowerful tools and starting points that already do the work, from a character sheet to product-into-UGC.
MCP & AI AgentConnect Claude or any other AI agent and let it run the studio while you do something else.Models
The right model for every job
Pick the model that is strongest at what you are making, and switch the moment a better one lands. One studio, one balance, no subscription per lab.
Nano Banana Pro
Studio-grade images with text that actually reads, up to 4K.
Learn moreVeo 3.1
Video with sound generated in the same pass, in 1080p or 4K.
Learn moreSeedance
Multi-shot video from text or an image, in 1080p.
Learn moreFLUX.2
Production-grade images up to 4MP, with up to ten reference images.
Learn moreNano Banana 2
Pro-level image generation and editing, at Flash speed.
Learn moreNano Banana
The original Gemini image model. Cheap, quick, good enough to explore with.
Learn moreKling 3.0
Cinematic video up to 15 seconds, with native audio and named elements.
Learn moreSeedream 5
Dense layouts and real typography, in fourteen languages.
Learn moreGPT Image 2
Near-perfect text in images, and it follows a long brief.
Learn moreMiniMax H3
A frontier video model that runs from 480p drafts up to 4K.
Learn moreIdeogram V3
Posters, logos and readable type, with three speed tiers.
Learn moreFeatures
Everything you need to ship real work
One good image is luck. Forty that belong together, on a deadline, is a studio.
Direct the shot, not the prompt
Pick a camera move, a lens and a light. The studio writes the prompt.
Describe it. The assistant builds it.
Say what you want in your own words. Name a model or let it pick one, and it sets the rest up.
One idea becomes a finished sequence
Describe your piece once and every shot comes back in the same style.
See it before you pay for it
Storyboard in stills first. Approve the frames, then shoot the video.
Your character, in every scene
One reference image, or your own LoRA, holds through the whole sequence.
Compare five models in one run
One prompt, five results side by side. Pick the winner instead of guessing.
Build it once, reuse it forever
Save how you work as a pipeline and run it for the next client unchanged.
Let an AI agent run the studio
Connect Claude over MCP and it generates while you do something else.
Know the price before you press go
The cost sits on the button. A run that fails costs you nothing.
Nothing you make gets lost
Everything lands in your library, searchable by prompt and by model.
Start halfway, not from scratch
Open a template that already does the job. Change the prompt and run it.
Every format, first time right
Vertical, wide or square. The studio sets it up on the model you chose.
In the studio
Work the way you want. Same models, same price.
Type a prompt, describe the job, build a pipeline, or hand it to an agent. Nothing is locked behind the way in that you pick.
Guides
Get better results, faster
The model is rarely the problem, the prompt is. These guides show you what to say, which model to say it to, and what it costs.
Pricing
Pay for what you make. Nothing else.
No seat fee, no monthly minimum, and nothing expires. Buy credits and they stay yours. Every run shows its price on the button before you press it, and it updates the moment you change the model, the length or the resolution.
All prices exclude VAT.
You pay per generation, in credits. What a run costs is on the button before you start it, so nothing catches you out. If something breaks on our side or at the provider, you get everything back. If a model refuses your request after it has already started, you pay only what that attempt actually cost and the rest is returned.
FAQ
Questions people ask first
What is dGEN Visual Studio?
A studio for generating images and video with AI. It carries 85 models: 15 that make an image from a prompt, 16 that edit an existing image, and 54 for video. You can work three ways: a straightforward generator, an assistant that sets it up for you, or a node canvas for a workflow you save and rerun.
What does it cost?
You pay per generation, in credits. What a run costs is shown on the button before you start it, and it depends on the model, and for video also on the length, the resolution and whether sound is generated. There is no subscription yet; the studio is not open for sign-ups.
What happens if a generation fails? Do I lose the credits?
It depends on why it failed, because the two cases cost us differently. If something broke on our side or at the provider, the provider does not charge us and you get every credit back. If the model refused your request after it had already started, the provider does charge us for the attempt, so you get our margin back and the purchase price stays with us. You never pay more than the attempt actually cost. If the provider reports it charged nothing, you get everything back.
Can I use the results commercially?
Yes. We claim no ownership of what you generate and place no restriction of our own on commercial use. Every model in the studio is marked by its provider as licensed for commercial use, and we checked all of them rather than a sample.
Do I own the copyright in what I generate?
Probably not automatically, and anyone who promises you otherwise is selling certainty they do not have. Under Dutch and European law copyright requires a human creator, so material generated purely from a prompt may not be protected at all. You are free to use it, but you may not be able to stop someone else from using something similar. The more of your own creative work goes in, the stronger your position.
Join the waitlist
The studio is not open to everyone yet. Leave your name and email address and we will let you know the moment you can get in.
We use your address only to tell you the studio is open, and for nothing else. You are on the list only after you confirm the email.