WaveSpeedAI Seedream 5.0 Pro Image Generation/Editing

Generate and edit images using ByteDance's Seedream 5.0 Pro (API Preview) via the WaveSpeed AI platform. Three endpoints: text-to-image, editing with up to 10 reference images, and layer decomposition into a base image and transparent layers.

Setup

Install the open-source CLI once and sign in; the CLI stores the key, so never ask the user to paste an API key into the chat:

npm install -g @wavespeed/cli
wavespeed login          # opens https://wavespeed.ai/accesskey and stores the key
wavespeed status         # confirms you are signed in

For CI or one-off shells, WAVESPEED_API_KEY in the environment also works.

The same platform is exposed as an MCP server: hosted at https://api.wavespeed.ai/mcp (browser sign-in) or locally with npx -y @wavespeed/mcp@latest; every example below maps onto run_model with the same model id and input fields.

Quick Start

Text-to-Image

OUTPUT_URL=$(wavespeed run bytedance/seedream-v5.0-pro \
  -p "A minimalist coffee shop logo with clean typography" \
  --json | jq -r '.outputs[0]')

Image Editing

The images parameter accepts an array of image URLs (1-10 images). For local files use the @path marker; the CLI uploads them for you.

# Upload a local image to get a URL
OUTPUT_URL=$(wavespeed run bytedance/seedream-v5.0-pro/edit \
  -i images='["@./photo.png"]' \
  -p "Add warm sunset lighting and lens flare" \
  --json | jq -r '.outputs[0]')

Existing URLs work as-is:

OUTPUT_URL=$(wavespeed run bytedance/seedream-v5.0-pro/edit \
  -i images='["https://example.com/photo.jpg"]' \
  -p "Add warm sunset lighting and lens flare" \
  --json | jq -r '.outputs[0]')

Shortcut commands

wavespeed generate image "<prompt>" -m seedream-5-pro and wavespeed edit image "<prompt>" --image @./photo.png -m seedream-5-pro (MCP generate_image / edit_image with model: "seedream-5-pro") use the same model and pick the endpoint automatically. seedream-5-pro is also the default model for these shortcuts.

API Endpoints

Text-to-Image

Model ID: bytedance/seedream-v5.0-pro

Generate images from text prompts.

Parameters

Example

OUTPUT_URL=$(wavespeed run bytedance/seedream-v5.0-pro \
  -p "A movie poster for a sci-fi thriller with bold title text 'HORIZON' at the top" \
  -i aspect_ratio="2:3" \
  -i resolution="2k" \
  -i output_format="png" \
  --json | jq -r '.outputs[0]')

Image Editing

Model ID: bytedance/seedream-v5.0-pro/edit

Edit one image or generate from several reference images (up to 10).

Parameters

Example

OUTPUT_URL=$(wavespeed run bytedance/seedream-v5.0-pro/edit \
  -i images='["@./portrait.png"]' \
  -p "Transform into a vibrant pop art style with bold colors" \
  -i resolution="2k" \
  --json | jq -r '.outputs[0]')

Multi-Image Editing

OUTPUT_URL=$(wavespeed run bytedance/seedream-v5.0-pro/edit \
  -i images='["@./face.png", "@./scene.png"]' \
  -p "Place the person from the first image into the scene from the second image" \
  -i aspect_ratio="16:9" \
  --json | jq -r '.outputs[0]')

Layer Decomposition

Model ID: bytedance/seedream-v5.0-pro/layer-decomposition

Separate a single image into a base image and transparent image layers. Read every entry of outputs, not only outputs[0].

Parameters

Example

wavespeed run bytedance/seedream-v5.0-pro/layer-decomposition \
  -i image=@./poster.png \
  -i output_format="png" \
  --json | jq -r '.outputs[]'

Pricing

Estimates from wavespeed price (per run, USD). Prices may change; check wavespeed price <model-id> ... or the model pages: text-to-image, edit, layer decomposition.

Edit cost rises slightly with more reference images (4 references at 1k: $0.054).

Prompt Tips

• Put text that must appear in the image in quotes inside the prompt (e.g. a title 'HORIZON').
• Pick aspect_ratio up front for the target placement (9:16 for vertical, 21:9 for banners) instead of describing it in the prompt.
• With several references, say which image supplies what ("the person from the first image", "the background of the second image").
• Use prompt_optimization_mode=fast for quick drafts; keep standard for long or detailed prompts.
• 1.5k costs the same as 1k, so it is the cheapest way to get more pixels.

CLI tips

# Inspect the live input schema before running (fields, enums, defaults)
wavespeed run bytedance/seedream-v5.0-pro -h

# Estimate the cost (no generation, no charge)
wavespeed price bytedance/seedream-v5.0-pro -p "..." -i key=value

# Save outputs to disk instead of only printing URLs
wavespeed run bytedance/seedream-v5.0-pro -p "..." --json --download "./out/{index}.{ext}"

# Local files: prefix the path with @ and the CLI uploads it and passes the hosted URL
wavespeed run bytedance/seedream-v5.0-pro -i <field>=@./local-file.png --json

# Recover a result if the run was interrupted (the id is in the --json output)
wavespeed show <id>

run --json prints { id, model, prompt, outputs: [url, ...], saved: [path, ...], elapsed_ms, raw }. Read outputs[0] for the result URL.

A run that fails with a gateway or network error (502/503/504, connection reset, timeout) does NOT mean the task failed, and the submit request is never retried automatically, so re-running can create a duplicate paid task. Do not re-run immediately. If the error contains a task id (task_id / prediction id), keep checking it with wavespeed show <id> --json until its status is terminal (completed, failed, cancelled, timeout, deleted). If no id was returned, check recent predictions with wavespeed history --json (add --model <id>; default window is the last 24h) and re-run once only if none matches. A client-side wait timeout or an interrupted run does not mean the task failed: tasks can run for a long time, so do not give up after a fixed time.

Security constraints

• Never ask for the key in chat: wavespeed login handles auth; if wavespeed status says signed out, ask the user to run it.
• Local files only via @: bare paths are passed through untouched and the model will reject them. Only @-prefixed values upload.
• No arbitrary URL loading: only pass media URLs the user provided or that came back from a previous run.
• Input validation: only pass parameters documented above; confirm with wavespeed run <model> -h when unsure.