Bytedance Seedream V4.5 API Documentation

Bytedance Seedream V4.5 API Documentation

Playground

Seedream 4.5 is a next-gen text-to-image model optimized for typography—crisper text rendering, stronger prompt adherence, and up to 4K output for posters and brand visuals. Ready-to-use REST inference API, best performance, no cold starts, affordable pricing.

Features

Seedream 4.5 is ByteDance’s high-resolution image generation model, upgraded through large-scale training and architecture refinement. It is especially strong at typography, poster composition, and branded visuals, with clear text rendering and strong prompt adherence.

Model highlights

• Enhanced typography – Renders sharp, legible text for posters, logos, UI, and marketing layouts.
• Designer-level composition – Handles complex poster-style layouts with clear hierarchy (title, subtitle, body text, logos).
• Strong prompt adherence – Closely follows detailed descriptions for subjects, layout, and style.
• High-resolution output – Supports custom width/height subject to the limits in the current input schema.

Recommended use cases

• Poster, banner, and KV design with embedded text
• Brand visual, logo, and campaign asset creation
• E-commerce product imagery and hero shots
• Social media graphics where typography is part of the design
• Presentation, landing-page, and in-app visuals

Pricing

• $0.04 per generated image

How to use

1. Enter your prompt Describe the subject, composition, text elements (e.g., title / subtitle / tagline), and overall style.

2. Set size (width & height) Choose pixel dimensions for your image. The model supports custom sizes as long as the dimensions satisfy the current input schema and the model’s limits.

3. Run the job Click Run to generate the image, then refine your prompt or size for the next iteration.

Suggested resolutions

Example resolutions:

Check the current input schema before using a custom size.

Notes

• For text-heavy posters, slightly higher resolutions (e.g. 2048×2048 or above) give noticeably cleaner typography.
• Keep logos and key text explicitly described (e.g. “white all-caps title at the top, small gray subtitle below”).

Model Comparison on WaveSpeedAI

Use Seedream 4.5 together with other models, depending on your priorities:

• google/nano-banana-pro/text-to-image – Google’s Nano Banana Pro (Gemini 3.0 Pro Image family) offers another option for image generation; compare current prices and output quality for your task.

• Tongyi-MAI/Z-Image-Turbo (available as Z-Image on WaveSpeedAI) – Tongyi-MAI’s 6B, 8-step turbo model focuses on maximum speed and throughput while keeping photorealism and bilingual (EN/ZH) support.

• wavespeed-ai/flux-2-pro/text-to-image – FLUX.2 [pro] is a flagship, general-purpose model for cinematic quality and complex scenes, great when you need broad stylistic range beyond typography-heavy posters.

• /seedream-v4 – The previous Seedream generation, strong at high-resolution illustration and diverse styles; Seedream 4.5 builds on it with noticeably better text rendering and layout control for branding work.

Rule of thumb:

• Choose Seedream 4.5 for posters, brand layouts, and any text-heavy creative.
• Choose Nano Banana Pro or Z-Image-Turbo after comparing speed, price, and quality on your inputs.
• Choose FLUX.2 [pro] for cinematic, style-flexible hero shots.
• Choose Seedream V4 when you want its familiar look or need variety across illustration styles.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "size": "2048*2048"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/bytedance/seedream-v4.5" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

Result Request Parameters

Result Response Parameters