Qwen Image Edit 2511 API Documentation

Qwen Image Edit 2511 API Documentation

Playground

Qwen Image Edit 2511 is a major upgrade over 2509 for real-world image editing and design. It delivers stronger edit consistency, robust multi-person identity/pose consistency, built-in LoRA styles, enhanced industrial/product design, and improved geometric reasoning for structure-preserving edits. Built for stable production use with a ready-to-use REST API, no cold starts, and predictable pricing.

Features

Qwen-Image-Edit-2511 is a high-consistency, production-grade image editing model built on the Qwen-Image 20B (MMDiT) architecture, delivering stronger real-world edits, better identity preservation, and more reliable multi-subject control than earlier releases. It’s designed for fast, prompt-driven edits with stable composition, clean details, and commercial-ready output quality.

What’s new in 2511

• Stronger multi-person consistency Handles group photos and multi-subject scenes with better stability and fewer identity swaps.

• Integrated popular community LoRA styles Built-in style options for common community aesthetics without extra setup (availability depends on the endpoint).

• Better industrial & product editing Cleaner structure, surfaces, and product geometry for design mockups and marketing visuals.

• Reduced drift across edits Improved identity and subject consistency when making iterative or larger edits.

• Improved geometric reasoning More reliable structural transformations and shape-aware editing.

Core capabilities

• Dual-mode editing

• Appearance editing: add/remove/modify elements while keeping other regions visually consistent.

• Semantic editing: global style/pose/scene transformations that preserve intent while allowing broader pixel changes.

• Precise text editing (when applicable) Add, delete, or replace on-image text while keeping natural typography behavior (spacing, alignment, style).

• Style preservation Maintains lighting, palette, and overall look while applying targeted changes.

Best for

• Multi-person projects — group photos, team portraits, event shots
• Industrial & product design — product mockups, packaging tweaks, commercial comps
• Identity-preserving edits — portraits, characters, avatar refinement
• Design & marketing teams — fast iterations, brand-safe edits, localization visuals
• E-commerce & social — product cleanup, background updates, quick visual variations

Example prompts

• Multi-person: Add a third person matching the existing lighting and camera angle.
• Industrial: Convert this product into a clean technical blueprint view with construction lines.
• Identity: Keep the person’s facial features unchanged and replace the background with a modern office.
• Appearance: Add a latte cup in the top-right corner without changing anything else.
• Semantic: Restyle the scene as cyberpunk while keeping the brand logo and layout consistent.

Parameters

How to use

1. Add your base image as the first item in images (you should see a preview in the UI).
2. Optionally add 1–2 more reference images (maximum 3 total) to guide style, subject details, or composition.
3. Write a clear prompt describing the edit and constraints (examples: “keep face unchanged”, “keep pose”, “keep background”).
4. Run the model and review the result.
5. Iterate by tightening constraints and making one major edit per run for best consistency.

Supported output formats typically include JPG / PNG / WEBP (as exposed by the endpoint).

Pricing

• $0.03 per edited image

Note

If you’re using image URLs (instead of uploading locally), make sure they’re publicly accessible. If the URL is valid, the interface will display a preview before you run the job.

Related Models

• Qwen Image Edit — AI Image Editing & Inpainting — Prompt-driven image editing for object removal, background replacement, and inpainting with fast iterations and strong instruction following.

• Qwen Image Edit Plus — High-Fidelity Image Editing — Higher-quality image edits with cleaner edges, improved detail retention, and more stable results on complex scenes and textures.

• Google Nano Banana Pro (Edit) — Photoreal Image Editor — High-fidelity image editing optimized for photoreal results, accurate text rendering, and composition-preserving transformations for professional creatives.

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",
  "images": [
    "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg"
  ],
  "output_format": "jpeg"
}
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/wavespeed-ai/qwen-image/edit-2511" \
  -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