Seedream 5.0
Generate, edit, and break images into layers on Seedream 5.0, ByteDance's image model that thinks before it creates. Pick the version you want or describe the frame in plain language, and the agent picks the best-suited one and writes the prompt.
Seedream 5.0 Flash now on invideo
Faster takes, lower cost
Flash generates faster and costs less per image than Pro. Run twenty variations before you commit to one, and exploring stops being the expensive part.
Layers straight out of the generation
Flash splits a finished image into a background and up to 16 separate layers, each exported as a transparent PNG. Move the headline, swap the product, keep everything else.
Ten references in one brief
A product shot, a character, a palette, and a layout go in together, up to 10 images per request. Flash reads them as one instruction.
Up to 2K, in every frame shape
Outputs run at 1K, 1.5K, or 2K, from 21:9 banners to 9:16 vertical feeds. Transparent backgrounds come built in.
Find it on Flash, finish it anywhere
Flash is where a frame gets found. When a shot needs Lite's 4K output or Pro's precise editing, the agent runs the direction you landed on through that version.
Try nowSeedream 5.0 Lite vs Pro vs Flash
Seedream 5.0 comes in three versions, each built for a different stage of the job. On invideo, all three sit on the same roster.
Seedream 5.0 Lite: the thinking model
Lite searches the web for live context, works through physics and spatial logic before it draws, and outputs up to 4K. Use it when the image has to be right, not just good-looking.
Seedream 5.0 Pro: the design model
Pro edits by pointing: boxes, arrows, sketches, and hex codes. It splits any image into up to 20 layers and renders text natively in 14 languages. Use it when the frame needs precise changes.
Seedream 5.0 Flash: the fast model
Flash has the same resolution range and layer separation as Pro, faster and at a lower cost. Use it when you're still finding the shot.
What Seedream 5.0 does best
Images grounded in what's happening now
Ask for this weekend's match poster or this week's trending topic, and Lite searches the web before it draws. The frame reflects the news, not the model's cutoff date.
Reasoning before rendering
Lite works through physics, spatial relationships, and domain knowledge before it commits a single pixel. A labeled diagram gets its labels in the right places.
Edit by pointing, not re-prompting
Draw a box, drop an arrow, sketch a rough shape, paste a hex code. Pro changes the area you marked and leaves the rest of the frame alone.
Every element on its own layer
Pro breaks a flat image into up to 20 editable layers with transparent backgrounds. Swap the product, move the headline, keep the lighting.
Text that reads right in 14 languages
Posters, menus, packaging, and slides render with correct characters in 14 languages, Arabic right-to-left layouts included. One design, localized without a redraw.
Dense information, still readable
Infographics, flavor wheels, technical blueprints, slide layouts. Pro organizes a lot of information into a frame that still reads at a glance.
Photographic realism
Physical lighting, real shadows, skin with texture. Portraits and retouches keep the subject looking like themselves.
One look across a whole set
Lite returns a sequence of related images in a single request, with the same characters, products, and style throughout. A campaign lands as a set, not as separate attempts.
How Seedream 5.0 works with invideo agents
The agent routes each frame to the right version
Seedream 5.0 sits on a roster of 200+ models. The agent sends work to Lite when a frame needs live context or 4K, to Pro when it needs precise edits or dense text, and to Flash when you're still finding the shot.
You direct, the agent writes the prompt
Describe the frame the way you'd brief a designer. The agent attaches the right references, in the right order, and writes the prompt the way Seedream reads best.
Your brand rides along on every image
Brand colors, fonts, product shots, and characters are attached to every generation. Image forty matches image one.
Stills that turn into shots
A Seedream frame can open a video. The agent passes your image to a video model like Seedance or Veo and keeps the look intact.
You always stay in control
You set how much the agent does on its own: let it generate everything, ask before each image, or see every prompt before it runs.
Who is Seedream 5.0 for?
Brand and product marketers
Create product shots, key visuals, and localized ad variants from one product photo. Layers keep the product exactly itself while everything around it changes.
Designers and creative teams
Create posters, social tiles, slide layouts, and infographics where the text renders correctly the first time. Export the layers straight into your design file.
Performance Ads and UGC teams
Run twenty static variants on Flash, find the winner, then finish it on Pro. Testing cycles move at the speed of the brief.
Educators and course creators
Create diagrams, scientific illustrations, and explainer visuals where every label lands in the right place.
Filmmakers
Create storyboard frames, character sheets, location concepts, and first frames for video, all in one consistent look.
Helping creatives stay creative
Multiplayer mode
Collaborate in real time with live cursors to show what everyone's working on.
Storyboarding
Turn any script or idea into a shot-by-shot plan, then tweak as needed before generating.
Script writing
Write your script inside invideo, and ask an AI co-writer for help if you'd like.
Timeline editor
Picture Premiere Pro with full AI.
Build your own agents
Create custom agents to fill specific roles like cinematographer, music designer, and more.
From solo creatives to creative enterprises
World-class investors stand behind invideo.
Backed by the firms behind Stripe, Spotify, Flipkart, and ByteDance.
Loading pricing…
Seedream 5.0 FAQs
What is Seedream 5.0?
Seedream 5.0 is ByteDance's image generation model, released in three versions. Lite (February 2026) adds live web search, reasoning, and up to 4K output. Pro (July 2026) adds precision editing, layer separation, and text in 14 languages. Flash (September 2026) is the faster, lower-cost version.
What's the difference between Seedream 5.0 Lite, Pro, and Flash?
Lite reasons, searches the web, and outputs up to 4K. Pro edits by pointing, splits images into up to 20 layers, and renders text in 14 languages, at up to 2K. Flash matches Pro's 2K range and layer separation, and generates faster at a lower cost.
What is Seedream 5.0 Flash?
Flash is the newest version of Seedream 5.0, built for speed and cost. It takes up to 10 reference images, outputs at up to 2K, and separates images into layers. It's on invideo now.
Which version should I use?
On invideo, you don't have to pick. The agent routes each frame to the version that wins it, and you can set your own routing rules for a project.
What resolution does Seedream 5.0 support?
Lite outputs up to 4K. Pro and Flash output at 1K, 1.5K, or 2K.
How many reference images can I use?
Up to 14 on Lite, and up to 10 on Pro and Flash.
Can Seedream 5.0 separate an image into layers?
Yes, on Pro and Flash. Pro splits an image into up to 20 layers, each with a transparent background, so you can edit one element without touching the rest.
Can Seedream 5.0 render text in other languages?
Yes. Pro renders text natively in 14 languages, including Arabic, Japanese, Korean, Russian, and Thai, with right-to-left layout support.
Can Seedream 5.0 search the web?
Lite can. When a prompt mentions a current event or trend, Lite pulls live context from the web before it generates.
Do I need to write prompts for Seedream 5.0 on invideo?
No. You describe the frame in plain language. The agent picks the version, attaches your references, and writes the prompt Seedream reads best.