seedream-4 - ByteSpike

seedream-4 is ByteDance’s flagship text-to-image and image-to-image model. The reason to reach for it over gpt-image-2: markedly stronger on Chinese in-image text, poster composition, and East-Asian product photography aesthetics, at meaningfully lower per-image cost. The reason to look elsewhere: Western Mediterranean / Northern European aesthetic still tilts toward the OpenAI / Google models. Pricing: $0.025 / image — see the rate card. Failures don’t bill; pricing sits between nano-banana (cheaper) and gpt-image-2 (pricier).

Protocols

Protocol	Path
OpenAI Images	`POST https://llm.bytespike.ai/v1/images/generations`

Quickstart

curl https://llm.bytespike.ai/v1/images/generations \
  -H "Authorization: Bearer $BYTESPIKE_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "seedream-4",
    "prompt": "中国风海报，水墨意境，远山近水，留白构图，标题：春日新品",
    "n": 1,
    "size": "1024x1792"
  }'

Capabilities

Capability	Supported
Text-to-image	✅
Image-to-image (with `source_image`)	✅
`n` ≥ 2 batch generation	✅
`size` 1024×1024 / 1024×1792 / 1792×1024	✅
Chinese in-image typography	✅ strong
`response_format` url / b64_json	✅
Modality	image
Capability bucket	`image_generate`

When to use

Chinese-market marketing — posters, product cards, social cards where Chinese text rendering and East-Asian aesthetic carry weight.
Mid-tier cost with image-to-image — better than nano-banana-v2 on hard prompts, cheaper than gpt-image-2.

When not to use:

You need the cheapest tier — try seedream-v5lite.
You need the absolute highest fidelity, regardless of price — seedream-4.5 or gpt-image-2.

seedream-4.5 — successor with refinements
seedream-v5lite — cheapest tier
gpt-image-2 — OpenAI alternative

gpt-image-2 seedream-4.5

​Protocols

​Quickstart

​Capabilities

​When to use

​Next

Protocols

Quickstart

Capabilities

When to use

Next