Flux 2 Flash Edit API Documentation

Flux 2 Flash Edit API Documentation

Playground

FLUX 2 flash Edit from Black Forest Labs delivers ultra-fast, production-grade image-to-image editing. Use plain-English prompts and precise HEX color control to restyle scenes, replace backgrounds, inpaint/outpaint, and retouch with brand-consistent palettes. Low-latency REST API, no cold starts, affordable pricing.

Features

FLUX.2 Flash Edit is a fast, cost-efficient image-to-image editing model for prompt-driven transformations. Give it one or more input images plus a clear instruction, and it will apply targeted edits while keeping the overall composition stable—great for high-throughput creative pipelines.

It’s best suited for quick style shifts, background tweaks, light retouching, and generating multiple variations for iteration or A/B testing.

Key capabilities

• Ultra-fast prompt-driven editing Apply natural-language edit instructions with low latency for rapid iteration.

• Composition-stable transformations Keeps layout and key identity cues stable while applying targeted changes.

• Multi-image editing (up to 3 inputs) Accepts multiple input image URLs to guide the edit and/or provide references.

• Production-friendly controls Deterministic results with seed, plus API-only options for synchronous responses and base64 outputs.

• Size control for common creative formats Generate outputs sized for banners, posters, social crops, and product imagery.

Parameters and how to use

• prompt: (required) The positive prompt describing what to change and what to preserve.
• images: (required) Array of input image URLs for editing (1–3 items).
• size: Output size as a string in width*height pixels (each dimension 256–1536).
• seed: Random seed (-1 for random; fixed value for repeatability).
• enable_base64_output: If true, returns the output as a base64 string instead of a URL (API-only).
• enable_sync_mode: If true, waits for generation to finish and returns the result in the same response (API-only).

Prompt

Write prompts like an editor’s brief: what to change, what to keep, and what “done” looks like.

Tips that work well for FLUX.2 Flash Edit:

• Start with the exact change: “replace the background with…”, “restyle as…”, “remove…”, “retouch…”
• Explicitly preserve what matters: “keep the same face, pose, framing, and logo text.”
• Keep edits scoped. If the output drifts, reduce the number of simultaneous changes and iterate in smaller steps.

Media (Images)

• images must be 1–3 URLs pointing to your input images.
• If your images are local files, upload them first (for example via WaveSpeedAI’s media upload endpoint) and use the returned download_url values here.
• For best stability, use clear, high-quality inputs and avoid heavily compressed thumbnails.

Other parameters

• size Use width*height (for example 1024*1024, 832*1216, 1216*832). Each dimension must be 256–1536.

• seed

• -1 (default behavior) for a new variation each run

• A fixed integer (for example 12345) for repeatable results and controlled iteration

• enable_sync_mode (API-only)

• false (default): submit a task, then query the prediction result

• true: wait and return the finished output in the same response (simpler integration, but the request takes longer)

• enable_base64_output (API-only)

• false (default): outputs are URLs

• true: outputs are base64 strings (useful when you can’t fetch remote URLs)

After you finish configuring the parameters, click Run, preview the result, and iterate if needed.

Pricing

Pricing starts at $0.03 per request, including one reference image. Each additional reference image costs $0.005.

Notes

• If you need consistency across a batch, keep size and seed fixed and vary only the prompt.
• When using multiple input images, describe how they should be used (for example: “use image 1 as the base; use image 2 as style reference”).
• For UI-safe outputs, keep instructions specific (avoid “make it more provocative” style prompts) and review results before publishing.

Related Models

• FLUX.2 Flash Text-to-Image – The same “flash” tier optimized for fast, low-cost generation from text prompts.
• FLUX.2 Dev Edit – A lightweight edit variant that’s useful when you want a different speed/cost/quality trade-off.
• FLUX.2 Flex Edit – A stronger choice when you need more controlled, style-consistent edits.
• FLUX 2 Max Edit – Higher-fidelity editing for detail-critical or “hero” assets.

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"
  ]
}
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/flux-2-flash/edit" \
  -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