IMAGE STUDIO
The best image models. One prompt box.
Seedream 5.0, GPT Image 2, Flux 2, Wan, Grok, Gemini — generate with any of them, edit your own photos image-to-image, and keep every result in your team’s workspace, not a downloads folder.
THE MODELS
Six providers. Zero tab-switching.
Every model routes through the same prompt box and the same library. Pick explicitly when you care, or leave it on Auto and let Teamday choose.
Seedream 5.0
ByteDance ModelArk
Photoreal detail, clean in-image typography, reference-image support
GPT Image 2
OpenAI
Instruction-following edits — powers image-to-image with up to 4 references
Flux 2 Flex
Black Forest Labs · FAL
Photoreal generalist with prompt expansion
Wan 2.7 / Qwen Image 2.0
Alibaba
Strong at readable text inside images
Grok Imagine
xAI
Fast image tier with exact billed-cost reporting
Gemini 3.1 Flash Image
Low-latency 1K output, good instruction following
IMAGE-TO-IMAGE
Edit yourimages, don't re-roll the dice
Start from your image
Upload a product photo, a character sheet, or a brand shot and describe the change. Editing runs through OpenAI’s image-edit endpoint or Seedream’s reference input — your subject stays your subject.
Up to 4 references at once
Combine a composition reference with a product reference in one generation. That is how Teamday ads keep the exact product in a brand-new scene.
Every aspect ratio that matters
Square 1:1, widescreen 16:9, vertical 9:16 — sized correctly per model (up to 2560×1440 on Seedream), not cropped after the fact.
WHERE IT LIVES
Not a toy in a tab — a tool in your company
Image generation is wired through the whole workspace, so the same engine serves quick one-offs, agent work, and full productions.
A dedicated generator
A clean prompt box with model and aspect pickers. Type, generate, compare, keep. Every result lands in your workspace library — nothing disappears into a chat scroll.
Inside chat, via your agents
Any Teamday agent can generate images mid-task: a marketer drafting social visuals, a designer exploring options. Ask in plain English; the asset arrives as a job with full provenance.
Inside the movie studio
The same engine renders storyboard keyframes for films and ads — chained image-to-image for scene continuity, with a product reference kept exact across every frame.
Turning stills into motion? See the AI video maker →
FAQ
Questions, answered
Which image models can I use?
ByteDance Seedream 5.0 and 4.5, OpenAI GPT Image 2, Black Forest Labs Flux 2 (via FAL), Alibaba Wan 2.7 and Qwen Image 2.0, xAI Grok Imagine, and Google Gemini 3.1 Flash Image. Pick one explicitly or leave the picker on Auto.
Can it edit an existing image, not just generate new ones?
Yes. Supply a reference image and a plain-English instruction. Edits run image-to-image through OpenAI’s images/edits endpoint or Seedream’s reference input, and you can pass up to 4 reference images in one generation.
Do I need my own API keys?
You can bring your own provider keys for full control, or use Teamday’s built-in platform keys for image generation on OpenAI, ModelArk, and FAL within a daily quota. Video, music, and speech are bring-your-own-key.
Where do generated images go?
Every generation is a tracked job whose result lands in your workspace media library — with the prompt, model, and provenance attached. Reuse any image as an agent avatar, a storyboard keyframe, or a reference for the next generation.
Can my AI employees generate images on their own?
Yes — agents have a generate_asset tool. A marketing agent can draft campaign visuals inside its chat or scheduled mission, and the results show up as reviewable jobs, not silent side effects.
Your next visual is one sentence away.
Generate with the best models, edit with references, and keep everything where your team works.
Start free trial