Skip to main content
AI image model
ByteDance
Seedream

Seedream 3.0

Seedream 3.0 / Capabilities — Volcengine has separately announced that the legacy LAS operator for `Doubao-Seedream-3.0-t2i`, version `250415`, ended service on 2026-05-11.

Online use is currently unavailable

This page is for reviewing the model's capabilities, use cases, and public references. The model does not appear in the home workbench, and you cannot submit generation tasks with it.

Capabilities
Text-to-image
Current price
No task creation
Prompt limit
5,000 characters
Upload limit
No image uploads
30-second overview
  • Seedream 3.0 / Current pricing — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.
  • Seedream 3.0 / Upload limit — Kyeo accepts no images and exposes no image-editing workflow.
  • Seedream 3.0 / Prompt adherence — Image_size includes square, square_hd, portrait_4_3, portrait_16_9, landscape_4_3, and landscape_16_9. Guidance_scale ranges from 1–10 and defaults to 2.5.
  • Seedream 3.0 — Volcengine has separately announced that the legacy LAS operator for `Doubao-Seedream-3.0-t2i`, version `250415`, ended service on 2026-05-11.

Popular searches: which Seedream 3.0 controls are available · why Seedream 3.0 is text-to-image only · what the Seedream 3.0 guidance_scale field means

Examples

Visual concept

Visual concept

Hyper-detailed 1:1 square composition featuring High-fashion editorial portrait of an elegant Korean woman wearing avant-garde couture in vivid cyan, ruby red, sunshine yellow, and jet black. The background transforms into a massive architectural installation made from folded origami-inspired geometric surfaces. Hidden within different folded planes are monochrome fashion portraits of the same woman viewed from multiple perspectives. The colorful main portrait stands confidently in the center with dramatic editorial lighting, premium fashion styling, luxurious textures, and clean minimal sophistication. Inspired by futuristic gallery architecture and luxury fashion campaigns. VERY COLORFUL No typography, no letters, no logo, watermark, symbols, captions, magazine title, or text.

Visual concept

Visual concept

Use the uploaded portrait only as the identity reference. Preserve the person’s real facial features, face shape, skin tone, and recognizable hairstyle. Do not copy the original outfit, background, pose, lighting, or image quality from the uploaded portrait. Create a realistic cinematic portrait of the same person sitting quietly on an old green-tiled rooftop under dense tree shadows. The person is relaxed, looking slightly downward with a soft calm expression. Surround the scene with dark green foliage, hanging red pomegranate fruits, scattered sunlight, and natural leaf-shaped shadows across the face, body, and roof tiles. Add a vintage handheld camera placed casually nearby as a storytelling detail. Choose an outfit that best matches the person’s appearance and personality, different from the scene inspiration: natural, clean, elegant, lightweight, and suitable for a dreamy outdoor summer portrait. Keep the styling tasteful and realistic. Low-angle composition, full-body or three-quarter body framing, cinematic natural light, strong contrast between deep shadow and bright sun patches, soft film grain, subtle lens flare, dreamy atmosphere, realistic skin texture, natural body proportions, authentic photography, shallow depth of field, poetic summer mood, high-detail editorial photo, no text, no logo, no watermark. Negative Prompt: watermark, logo, text, Chinese characters, AI-generated mark, over-smoothed skin, plastic skin, fake face, anime style, cartoon style, 3D render, doll-like face, distorted body, bad anatomy, extra fingers, missing fingers, deformed hands, deformed feet, unnatural legs, awkward pose, blurry face, low resolution, harsh filter, over-sharpened, messy composition, duplicated person, wrong identity, copied outfit from input portrait, copied background from input portrait.

Visual concept

Visual concept

Ultra-realistic fashion lifestyle photography of a cheerful young woman with shoulder-length wavy dark brown hair, wearing round thin-metal eyeglasses, small hoop earrings, and subtle natural makeup with bright orange lipstick. She is seated comfortably in a modern molded lounge chair with a light-blue cushioned interior and smooth curved design. Her pose is relaxed and confident, with one hand gently touching her glasses and the other resting naturally on the chair armrest. She is smiling directly at the camera with a warm, genuine expression. She wears a stylish tailored orange jumpsuit featuring bold white abstract geometric line patterns, a belted waist, rolled-up sleeves, wide-leg pants, and a soft satin-like fabric with realistic folds and texture. The outfit maintains the exact same pattern and design, only changing the original red color to a vibrant orange. She pairs it with clean white crew socks and minimalist white leather sneakers. Studio fashion editorial setup with a seamless vibrant blue background and matching blue wall, contrasted by a smooth orange floor. Full-body composition, centered framing, seated pose, legs casually crossed, modern contemporary aesthetic. Soft professional studio lighting with balanced highlights and shadows, sharp focus, highly detailed skin texture, realistic fabric rendering, premium fashion campaign photography, magazine-quality styling, natural color grading, ultra-realistic, DSLR quality, shallow depth of field, crisp details, luxury commercial photoshoot, 8K resolution. Negative Prompt: blurry, low resolution, extra fingers, deformed hands, distorted face, bad anatomy, duplicated limbs, unrealistic proportions, overexposed lighting, oversaturated colors, watermark, logo, text, cropped feet, cartoon, CGI, illustration, painting, artificial skin texture, exaggerated makeup, unnatural pose.

Visual concept

Visual concept

Create image: use the attached image as the main facial reference while preserving recognizable identity, natural facial proportions, facial harmony, realistic skin texture, visible pores, and natural asymmetry. Create a premium close-up studio beauty portrait with the locked dreamy vintage coquette rendering. The subject is photographed in a softly angled three-quarter composition with the shoulders gently turned while the face remains almost square to the camera. Her expression is calm, elegant, and quietly confident with naturally relaxed half-lidded eyes, softly parted lips, and a refined elongated neck. The composition remains intimate and face-dominant with elegant negative space above the head. Her ash chocolate-brown hair is styled into a soft romantic French twist inspired updo with airy crown volume, naturally separated strands, wispy face-framing tendrils around the temples and jawline, delicate baby hairs, and realistic strand separation. The hairstyle feels effortless, lightweight, and softly imperfect rather than polished or tightly sculpted. (Make up): luxury fresh glass-skin makeup with hydrated luminous skin, clearly visible realistic pores and continuous natural microtexture, brushed up feathered brows, muted olive-green contact lenses, long individually separated lashes, diffused dusty old-rose and mauve eyeshadow with a soft champagne shimmer concentrated near the inner corners, softly blended old-rose blush across the cheeks, localized champagne highlights across the cheekbones, brow bone, nose bridge, and Cupid's bow, and glossy muted rose-beige lips with realistic lip texture and natural lip lines. She wears sculptural polished vintage gold knot earrings and a dusty blush-pink chiffon couture blouse featuring finely pleated fabric, a softly gathered high ruffled neckline, delicate texture, and elegant minimalist tailoring. Photograph using a premium professional studio setup with an 85mm portrait-lens perspective and shallow depth of field. Keep the eyes, eyelashes, lips, freckles, skin texture, hair strands, pleated fabric, and jewelry extremely crisp while allowing the background to dissolve into smooth natural softness. Use soft off-camera diffused beauty lighting wrapping gently from the upper left, creating delicate dimensional facial shadows, creamy highlight roll-off, restrained contrast, and true-to-life skin volume. Maintain the locked rendering with natural color reproduction and absolutely no cinematic grading. Color rendering follows the locked coquette workflow with a muted old-rose palette instead of orange or golden tones. Whites remain soft ivory, highlights carry a delicate dusty blush tint, shadows remain pale taupe and warm greige, skin retains natural peach-beige warmth balanced with subtle old-rose reflections, and the entire photograph feels softly faded, powdery, elegant, and timeless. The background is a seamless warm muted beige studio backdrop with dusty greige, faded blush, and soft old-rose undertones. The matte surface contains only subtle tonal transitions with no visible backdrop edge, curtain, octabox, softbox, reflector, light panel, stand, cable, or studio equipment. Elegant off-center composition with refined negative space, premium editorial beauty photography, dreamy vintage coquette atmosphere, crisp optical clarity, realistic material rendering, creamy highlight roll-off, and true-to-life photorealism. Aspect ratio 4:5. Negative prompt: altered identity, narrow or elongated face, hollow cheeks, orange color grading, golden cinematic tones, amber skin, yellow cast, harsh HDR, oversaturated colors, plastic skin, waxy skin, excessive smoothing, oily skin, blurry face, soft eyes, exaggerated contour, oversized lips, distorted anatomy, CGI, illustration, anime, doll face, visible studio equipment, backdrop edge, text, watermark, logo.

Visual concept

Visual concept

{ "REFERENCE_IMAGE": "[Reference Image]", "MAIN_HEADLINE": "[Main Headline]", "LEFT_TEXT": "[Left Text]", "RIGHT_TEXT": "[Right Text]", "FOOTER_TEXT": "[Footer Text]", "BACKGROUND_COLOR": "[Background Color]", "prompt": "Create a high-end editorial fashion poster inspired by luxury magazine covers and premium advertising campaigns. Preserve the identity of the subject from the [Reference Image] while matching the overall artistic style. Use a dramatic close-up portrait composition with the subject positioned slightly off-center. Capture the subject from a subtle low angle to create a powerful, confident appearance. Add a strong radial motion blur effect extending outward from the center of the image while keeping the subject's face sharp and highly detailed. Place [Main Headline] as one continuous handwritten brushstroke phrase flowing organically across the entire composition. Allow some portions of the lettering to appear behind the subject while other portions overlap the foreground. Add [Left Text] and [Right Text] as small uppercase editorial typography blocks with generous spacing. Place [Footer Text] subtly in one corner. Style details: - Oversized fashion sunglasses - Minimal jewelry - Natural skin texture - Wind-swept hair - Luxury fashion styling - Elegant pose - Cinematic color grading - Fine film grain - Soft highlights and deep shadows - Premium studio photography quality Background: - Clean monochromatic backdrop - Use [Background Color] as the dominant color - Dynamic zoom-blur effect - Soft glow and light bloom Visual style: - Vogue-inspired editorial design - Luxury advertising aesthetic - Photorealistic rendering - Premium typography composition - High-fashion magazine cover design Negative prompt: - watermark - logo - blurry face - distorted anatomy - duplicated accessories - plastic skin - excessive sharpening - oversaturated colors - low quality output", "aspect_ratio": "4:5" }

Visual concept

Visual concept

Minimalist luxury watch campaign, oversized wristwatch positioned like modern sculpture, model leaning against the dial, giant word “TIME” in bold typography behind, deep emerald studio background, reflective polished floor, Swiss luxury advertising aesthetic, cinematic editorial styling, ultra-clean composition, 1:1

At a glance

Is this model right for you?

Integration — Volcengine has separately announced that the legacy LAS operator for `Doubao-Seedream-3.0-t2i`, version `250415`, ended service on 2026-05-11.

What it does best
Seedream 3.0 / Current price — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.
Who it is for
Seedream 3.0 — Do not choose it when the task requires image uploads or editing.
Why use it on Kyeo AI
Text-to-image with no uploads / Currently disabled — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

Key facts

Quickly evaluate whether this model fits your business use case.

Category
AI image model
Vendor
ByteDance
Model family
Seedream
Integration
Site runtime disabled
Runtime
No task creation
Prompt limit
5,000 characters
Upload limit
Submission unavailable

Other names

The same model may have different names in official documentation, community discussions, and API integrations. This section brings them together for you.

Seedream 3.0
Seedream v3
Seedream 3.0 API

Common questions

These practical questions focus on the task, input conditions, and result requirements you should confirm before choosing.

which Seedream 3.0 controls are available
why Seedream 3.0 is text-to-image only
what the Seedream 3.0 guidance_scale field means
which Seedream 3.0 image_size options are available
what range Seedream 3.0 accepts for seed
Seedream 3.0 vs Seedream 4.0
Seedream 3.0 vs Seedream 5.0 Lite

Decision guide

Still deciding? Start with these key decision points.

1
Disabled interface-reference page · Seedream 3.0

Disabled interface-reference page · Seedream 3.0 — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

2
Use another model for image input or the current official line · Seedream 3.0

Use another model for image input or the current official line · Seedream 3.0 — Compare Kyeo's Seedream 4.0, 4.5, or 5.0 Lite when references or editing are required. Volcengine has ended its legacy 3.0 LAS operator, so new workflows should also assess newer available models.

3
No generation controls are currently displayed · Seedream 3.0

No generation controls are currently displayed · Seedream 3.0 — Volcengine has separately announced that the legacy LAS operator for `Doubao-Seedream-3.0-t2i`, version `250415`, ended service on 2026-05-11.

Popular comparisons

Compare common alternatives on the same task to clarify differences in inputs, controls, and cost.

What is the difference between Seedream 3.0 and Seedream 4.5?
Seedream 3.0 vs Seedream 4.5
What is the difference between Seedream 3.0 and Nano Banana Pro?
Seedream 3.0 vs Nano Banana Pro
What is the difference between Seedream 3.0 and Flux 2 Pro?
Seedream 3.0 vs Flux 2 Pro

Model comparison table

Not sure which model to choose? Compare the key differences between this model and its alternatives at a glance.

Workflow scope
Seedream 3.0
Workflow scope — Text-to-Image only
Seedream 4.5
Workflow scope — Seedream 4.5: A currently available alternative in the same category; open its detail page to compare capabilities and pricing.
Seedream 5.0 Lite
Text-to-Image and image editing
Key controls
Seedream 3.0
Key controls — image_size, guidance_scale, seed
Seedream 4.5
Key controls — Seedream 4.5: A currently available alternative in the same category; open its detail page to compare capabilities and pricing.
Prompt limit
Seedream 3.0
Prompt limit — 5,000 characters
Seedream 4.5
3,000 characters
Seedream 5.0 Lite
3,000 characters
Current pricing
Seedream 3.0
Current pricing — No current task creation or credit deduction
Seedream 4.5
Current pricing — Seedream 4.5: A currently available alternative in the same category; open its detail page to compare capabilities and pricing.
Seedream 5.0 Lite
6 credits per request
Service status
Seedream 3.0
Service status — Disabled interface-reference page
Seedream 4.5
Service status — Seedream 4.5: A currently available alternative in the same category; open its detail page to compare capabilities and pricing.
Seedream 5.0 Lite
Kyeo generation and editing entry

Real-world experience

Practical guidance based on public sources, current on-site limits, and representative tasks.

Seedream 3.0 · Use another model for image input or the current official line

Seedream 3.0 · Use another model for image input or the current official line — Volcengine has separately announced that the legacy LAS operator for `Doubao-Seedream-3.0-t2i`, version `250415`, ended service on 2026-05-11.

Seedream 3.0 · Legacy numeric controls do not guarantee stable reproduction

Seedream 3.0 · Legacy numeric controls do not guarantee stable reproduction — Using the same parameters does not guarantee reproduction across time, tasks, or versions.

Capabilities

Text-to-image with no uploads

Text-to-image with no uploads — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

6 size identifiers and guidance_scale

6 size identifiers and guidance_scale — Image_size includes square, square_hd, portrait_4_3, portrait_16_9, landscape_4_3, and landscape_16_9. Guidance_scale ranges from 1–10 and defaults to 2.5.

Legacy numeric controls do not guarantee stable reproduction

Legacy numeric controls do not guarantee stable reproduction — Using the same parameters does not guarantee reproduction across time, tasks, or versions.

Use cases

Seedream 3.0 / Disabled interface-reference page

Seedream 3.0 / Disabled interface-reference page — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

Seedream 3.0 / Use another model for image input or the current official line

Seedream 3.0 / Use another model for image input or the current official line — Compare Kyeo's Seedream 4.0, 4.5, or 5.0 Lite when references or editing are required. Volcengine has ended its legacy 3.0 LAS operator, so new workflows should also assess newer available models.

Seedream 3.0 / Legacy numeric controls do not guarantee stable reproduction

Seedream 3.0 / Legacy numeric controls do not guarantee stable reproduction — Using the same parameters does not guarantee reproduction across time, tasks, or versions.

Prompt tips

Disabled interface-reference page / Seedream 3.0

Disabled interface-reference page / Seedream 3.0 — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

Use another model for image input or the current official line / Seedream 3.0

Use another model for image input or the current official line / Seedream 3.0 — Compare Kyeo's Seedream 4.0, 4.5, or 5.0 Lite when references or editing are required. Volcengine has ended its legacy 3.0 LAS operator, so new workflows should also assess newer available models.

Legacy numeric controls do not guarantee stable reproduction / Seedream 3.0

Legacy numeric controls do not guarantee stable reproduction / Seedream 3.0 — Using the same parameters does not guarantee reproduction across time, tasks, or versions.

Why choose it

Capabilities — The text-to-image boundary is explicit and does not imply editing support.
Prompt adherence / 6 size identifiers and guidance_scale — Image_size includes square, square_hd, portrait_4_3, portrait_16_9, landscape_4_3, and landscape_16_9. Guidance_scale ranges from 1–10 and defaults to 2.5.
Current price — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

What to know first

Seedream 3.0 / Currently disabled — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.
Seedream 3.0 / Integration — Volcengine ended the corresponding legacy 3.0 LAS operator on 2026-05-11.
Seedream 3.0 / Prompt limit — Complex text layout and tasks requiring absolute consistency still need small-sample validation.
Seedream 3.0 / Upload limit / Currently disabled — Kyeo accepts no images for this entry, so it cannot perform reference editing or local changes.
Seedream 3.0 / Legacy numeric controls do not guarantee stable reproduction — Text, composition, and repeatability all require practical validation.

Currently unavailable

This page is for reviewing the model's capabilities, use cases, and public references. The model does not appear in the home workbench, and you cannot submit generation tasks with it. Seedream 3.0 / Current price — The site runtime is currently disabled, with no generation entry, task creation, or credit deduction.

Image size
Optional
Parameter type: Select
Default: square_hd
Square
Square HD
Portrait 4:3
Portrait 16:9
Landscape 4:3
Landscape 16:9
Prompt adherence
Optional
Parameter type: Number
Default: 2.5
Range 1 - 10
Random seed
Optional
Parameter type: Number
Range 0 - 2147483647

Frequently asked questions

Related models

Still deciding? Take a quick look at these similar alternatives.

Seedream 4.5

Seedream 4.5 / Capabilities — ByteDance's official Seedream 4.5 material highlights multi-image editing, preservation of reference details, layout, and text rendering. The Kyeo workflow has its own concrete limits: prompts can contain up to 3,000 characters, 8 aspect ratios are available, and editing accepts up to 14 JPG, PNG, or WebP images at 10 MB each. Kyeo AI currently charges a fixed 7 credits per request, with the same price for basic and high. Budget for retries, revisions, and the share of outputs that pass review.

AI image model
7 credits per request

Nano Banana Pro

Nano Banana Pro supports text-to-image generation and reference-image editing on Kyeo AI, with 1K, 2K, or 4K output. Each 1K or 2K request costs 8 credits, while 4K costs 14. Resolution is an output-size tier, not a guarantee of quality, accuracy, or detail.

AI image generation and editing model
8–14 credits per request

Flux 2 Pro

Flux 2 Pro / Capabilities — Flux 2 Pro — Text-to-image and editing use the same resolution pricing: 5 credits for 1K or 7 credits for 2K. 3–5,000 characters; Up to 8 images.

AI image model
5–7 credits per request

Sources

This page is based on vendor materials, model-family documentation, and Kyeo AI integration settings.

Source note

Volcengine has separately announced that the legacy LAS operator for `Doubao-Seedream-3.0-t2i`, version `250415`, ended service on 2026-05-11.

Last updated: 2026-08-26
Volcengine: Legacy LAS operator service termination
Official
Open original source