Provider Seedream

Seedream

Mode: image · Models: 2

Vendor: BytePlus ModelArk · Official API docs: Seedream on ModelArk

Seedream (by ByteDance, on BytePlus ModelArk) is a high-fidelity text-to-image model with multi-image reference input and high-resolution output. The catalog exposes the flagship Seedream 4.5 (up to 4K) and the faster, lighter Seedream 5.0 Lite.

Models

id	Name	Input type
seedream-4.5	Seedream 4.5	t2i
seedream-5.0-lite	Seedream 5.0 Lite	t2i

CLI

# text-to-image at 4K
gen-ai generate -m seedream-4.5 \
  -p "a marble statue of a fox in a sunlit gallery, dramatic side lighting" \
  --ar 16:9 -r 4K

# multi-image reference (up to 2 source images), 4 variations
gen-ai generate -m seedream-4.5 -p "blend these into one cohesive product shot" \
  -i ./ref-a.jpg -i ./ref-b.jpg -n 4

# faster, lighter variant
gen-ai generate -m seedream-5.0-lite -p "minimalist poster, bold typography" --ar 3:4

MCP

{ "name": "picsart_generate",
  "arguments": {
    "model": "seedream-4.5",
    "prompt": "a marble statue of a fox in a sunlit gallery, dramatic side lighting",
    "aspectRatio": "16:9",
    "resolution": "4K",
    "count": 4
  } }

Parameters — seedream-4.5

Param	CLI flag	Type	Values
prompt	-p	text	required
resolution	-r	enum	2K · 4K (default 2K)
aspectRatio	--ar	enum	1:1 · 4:3 · 3:4 · 16:9 · 9:16 · 3:2 · 2:3 · 21:9 (default 16:9)
count	-n	enum	1 · 2 · 4 · 6 · 8 · 10 (default 1)
imageUrls	-i	file	up to 2 source/reference images
negativePrompt	--neg	text	elements to avoid

Source: gen-ai models info seedream-4.5 --json. Seedream 5.0 Lite shares the same surface; per the vendor worker its resolutions are gated to 2K/3K (4K rejected).

Pricing

gen-ai pricing seedream-4.5 -r 4K -n 4

Cost scales with resolution and the count of images generated.