Photos that feel unposed
Two diners in a neighborhood bistro, slight underexposure, high-ISO grain and a glass in the foreground: Seedream 5.5 keeps the light, linen and worn wood believable instead of glossy.

Seedream 5.5 is ByteDance's newest image model, built for realistic photos, clear small text, stable multi-turn edits and native transparent images. Describe a picture below to start.
Six Seedream 5.5 images and the prompts behind them. Hover or tap an image to read its prompt, then send it to the generator above.

Create a graffiti collage poster about collaborative street printing, using lime green, aubergine purple and mint over a gray-white photocopied paper texture. Oversized “PRINT PLAY!” lettering overlaps diagonally. A grayscale printmaker at the upper right holds a roller and print sample, with printing tools and handwritten phrases including “MAKE A MARK” and “PASS IT ON” below. Preserve matte halftone ink, fine grain and legible lettering in a layered, asymmetric composition.

A candid horizontal street photograph of a long-haired woman in a burgundy top and light brown jacket looking down at her phone. Strong wind sweeps her hair across her sunglasses and face. The entire frame tilts about 25 degrees clockwise, with consistent perspective across the figure, buildings, lamppost and pavement. Natural clothing folds, hurried framing and asymmetric cropping create an unposed everyday moment.

Create an illustrated infographic poster titled “FROM SEED TO SALAD” on warm-white paper with a handmade print texture. One continuous path snakes across three tiers, linking six numbered stages in order: SOW, SPROUT, GROW, TEND, HARVEST and SHARE. Five correctly directed arrows connect the sequence. Pair miniature gardening scenes and English descriptions with each stage, keeping the reading order clear, lettering legible and colors soft yet varied.

A horizontal ultra-wide-angle photograph of a curious gray-brown donkey peering through a vertical gap in a weathered wooden fence. Its broad muzzle nearly touches the lens, while its eyes and long ears recede in exaggerated close-up perspective. Focus on the muzzle’s fine hair, whiskers and skin texture, with softly blurred fence boards, blue sky and grass beyond. Natural daylight creates a playful candid moment.

A fantasy picture-book illustration with delicate gouache textures, set in a cloud-sea glider workshop. A bear mechanic steadies the mast, a red panda pilot pulls the sail rope from inside the boat, and an otter assistant approaches in the foreground carrying a wooden securing pin. A low, upward-looking view and gently tilted perspective emphasize the bow, triangular sail and tall timber roof. Cream, deep teal and honey-colored wood create a bright, warm scene of teamwork.

A hand lifts a string of water droplets from a miniature swimming pool. The droplets spell “seedream” vertically. Surreal summer photography with vivid colors and a film-like texture.
Each Seedream 5.5 edit starts from the previous result. Areas you did not touch stay pixel for pixel, with no color shift, so a painting, a product or a face survives many rounds.




Edit 1 / 5The paper, the vase and the first branch never move; new elements arrive in the same brush style. ByteDance pushed Seedream 5.5 to 100 successive edits of one portrait, still swapping rings and eyeglass frames precisely at the end.
Seedream 5.5 can return an image with a real alpha channel. Switch the backdrop and the same lettering, soft shadow included, sits cleanly on every color.

Prompt
Create “Seedream 5.5” inspired by the reference texture and style. Use rounded, bold sans-serif letters wrapped in alternating royal-blue and lake-blue yarn, with visible fibers, fuzz and side shadows. Use a 16:9 landscape format with a transparent background outside the lettering.
Seedream 5.5 writes alpha for clear areas and for semi-transparent ones such as shadows, glass, smoke or fur, with edges drawn by the model rather than cut out later.
Describe an asset, or point the model at a reference image to borrow its texture, as the yarn lettering above does.
Transparent stickers, characters and titles drop into posters, slides and mockups with no background removal step.
Seedream 5.5 is ByteDance Seed's next-generation image creation model, announced on November 9, 2026. It follows Seedream 5.0 Pro, creates images from a text prompt or reference images, and edits images you already have.
Seedream 5.5 brings three upgrades. Multi-turn editing is steadier: unedited regions are preserved at the pixel level with no color shift, across up to 100 rounds. Seedream 5.5 generates native RGBA images with full or partial transparency. And photos look more real, with natural light, materials and skin, while dense small text in English and Chinese comes out more accurately.
ByteDance says a Pro version of the Seedream 5.5 series will follow, without a date. Seedream 5.5 is already live in Dreamina by CapCut and on third-party API platforms.
At a glance
Seedream 5.5 makes photos believable and posters usable without a retouch pass.
Two diners in a neighborhood bistro, slight underexposure, high-ISO grain and a glass in the foreground: Seedream 5.5 keeps the light, linen and worn wood believable instead of glossy.

A French bakery poster sets "Pain au chocolat — 2,50 €" in elegant serif letters, with the accent, the dash and the euro sign in the right places.

A collectible-character board mixes the title BERRY CLUB with a Chinese tagline, character names and variant labels. Seedream 5.5 keeps the small bilingual type crisp and the layout orderly.

Built only from published facts. Seedream 5.5 quality numbers come from ByteDance's own tests.
| Feature | Seedream 5.5 | Seedream 5.0 Pro | GPT Image 2.6 |
|---|---|---|---|
| Developer | ByteDance Seed | ByteDance Seed | OpenAI |
| Released | Nov 9, 2026 | Previous generation | Nov 4, 2026 |
| Multi-turn editing | Pixel-level preservation, shown across 100 rounds | Reference-guided editing | Edits endpoint; Sunburst steadier over repeated edits |
| Transparent output | Native RGBA, full and partial | None; new in Seedream 5.5 | Not in the 2.6 release notes |
| Text rendering | 0.4% severe and 1.7% minor text corruption (ByteDance) | Higher rates than Seedream 5.5 (ByteDance) | OpenAI says typography can still slip |
| Realism | +55.62% GSB over 5.0 Pro, image to image (ByteDance) | Baseline | No published comparison |
| API price | From $0.02 per image (third party) | Varies by platform | $30 per 1M image output tokens |
In ByteDance's MagicArena tests across six production uses (e-commerce, marketing, film and TV, gaming, office work and fashion design), Seedream 5.5 beat 5.0 Pro on quality and usability in all six, and more than doubled usability in film and TV and in office work. ByteDance also reports that Seedream 5.5 leads GPT Image 2.5 Sunburst and Nano Banana 2.1 on text structure, with a clear lead in text-to-image realism.
ByteDance has not published official Seedream 5.5 pricing. Public third-party prices, checked November 10, 2026:
| Platform | What you get | Price |
|---|---|---|
| WaveSpeedAI | Seedream 5.5 text to image: 15 aspect ratios, 1K to 2K, transparent or opaque background | $0.02 at 1K and 1.5K, $0.027 at 2K |
| Atlas Cloud | Seedream 5.5 text to image, editing with 1–10 references, layer decomposition into up to 99 transparent layers | $0.02 per image |
| Volcano Engine Ark / BytePlus | Official Seedream 5.5 API | Not published yet |
| Dreamina by CapCut | Seedream 5.5 in the Dreamina app | In-app plans |
Here a Seedream 5.5 image costs 5 / 10 credits at 1K / 2K, with no API key and no separate bill. Extra reference images can add a little; the Generate button always shows the exact cost before you submit.
Write your Seedream 5.5 prompt in the box above: subject, setting, style, light and any quoted words.
Choose one of 7 aspect ratios and 1K or 2K. Sign in to add up to 10 reference images.
Press Generate and the Seedream 5.5 image starts in your workspace. To edit, attach the result as a reference and describe one change.
Seedream 5.5 follows plain, concrete language best, the way ByteDance writes its own examples.
Open with what the picture is about, then add setting, lens, light and style.
Put headlines and labels in quotes and say where they sit. Seedream 5.5 text is far more reliable, but proofread prices, dates and names.
For a cut-out, describe the subject alone and ask for a transparent background outside it.
Name the change and what must stay: "add a red seal, keep the vase unchanged". Small steps are where Seedream 5.5 multi-turn editing shines.
For a recurring character or product, attach the same reference images to every prompt.
Write a Seedream 5.5 prompt, pick a format and generate.
Start creating
Generate stunning AI images with Nano Banana. Fast, versatile, and free to start. Create beautiful artwork, photos, and designs with text prompts.

Professional AI image generation with Nano Banana Pro. Higher resolution, better detail, and advanced features for professional creators.

The latest Nano Banana 2 model. More affordable, supports up to 10 reference images, and delivers excellent quality at lower cost.

Fast and affordable AI image generation with Z Image. Text-only generation at just 4 credits per image. Perfect for quick iterations.
GPT Image 2 is OpenAI's next-gen image model for in-image text rendering, pixel-level editing, and world-knowledge-driven realism. Supports text-to-image and image-to-image with up to 16 reference images.
GPT Image 2.5 is OpenAI's newest image model, released September 8, 2026. Generate and edit images online with GPT Image 2.5 Flare for speed or GPT Image 2.5 Sunburst for precision: sharper detail, faithful reference photos, targeted edits that leave the rest untouched, and 1K / 2K / 4K output with up to 16 reference images.
GPT Image 2.6 is OpenAI's November 2026 image model with Flare and Sunburst tiers, complexity control, markup edits, SVG and GIF output. Compare the tiers, check pricing, and start creating.