Flux 2 Flex Edit API Documentation

Flux 2 Flex Edit API Documentation

Playground

FLUX.2 [flex] Edit delivers precise image-to-image editing from Black Forest Labs—apply natural-language instructions and exact hex color control for consistent, studio-quality results. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

FLUX.2 [flex] Edit is a configurable image editing model built on FLUX.2 [flex], aimed at teams that need fine control over how edits are applied. It can work with one or multiple reference images and lets you tune quality–speed trade-offs, making it a good fit for style-heavy, budget-conscious production workflows.

Act Well On

• Multi-image product and lifestyle compositions

• Brand asset refinement with style or reference images

• Typography and layout touch-ups in existing designs

• E-commerce and marketing visuals that need frequent updates

Why it looks great?

• Multi-image aware composition

Combine several reference images in a single edit when building product collages or style transfers. You can point to specific references by index (for example, “use the background from image 2 and the typography from image 4”) or simply describe them in the prompt.

• Adjustable quality and speed

Control inference steps based on how complex the edit is: quick colour swaps or minor clean-up can run with fewer steps, while dense multi-image edits can use higher settings for extra detail.

• Guidance control for edits

Decide how strictly the model should follow your instructions versus preserving the original look. Dial guidance lower for looser, creative reinterpretations; raise it when you want near-literal edits.

• Strong text and layout handling

Well suited for fixing signage, labels, and UI text inside images, keeping typography sharp while updating wording, colours, or layout.

• Natural language plus hex colour control

Describe edits in plain language (“make the jacket match the blue from our brand palette”) and use hex codes when you need exact corporate colours for products, UI elements, or backgrounds.

• LoRA- and pipeline-friendly

Works smoothly with LoRA adapters for brand or domain-specific styles, and its configurable nature makes it easy to slot into larger editing pipelines that need both flexibility and cost control.

• Output ready for production

Exports JPEG so edited assets can go straight into design tools, websites, or print-oriented workflows without extra conversion.

Pricing

Simple per-image billing:

• $0.12 per edited image

FLUX.2 family on WaveSpeedAI

Mix and match FLUX.2 models for a full generate-and-edit workflow:

• FLUX.2 [dev] Text-to-Image – lightweight base model optimised for speed and LoRA training.
• FLUX.2 [dev] Edit – fast, style-consistent edits on existing images with a lean architecture.
• FLUX.2 [flex] Text-to-Image – versatile, style-rich generation with broader aesthetics at high speed.
• FLUX.2 [pro] Text-to-Image – higher-capacity model for maximum-quality hero shots and demanding production work.
• FLUX.2 [pro] Edit – premium editing for detailed, high-fidelity transformations on critical assets.

More Image Tools on WaveSpeedAI

• Nano Banana Pro – Google’s Gemini-based text-to-image model for sharp, coherent, prompt-faithful visuals that work great for ads, keyframes, and product shots.
• Seedream V4 – ’s style-consistent, multi-image generator ideal for posters, campaigns, and large batches of on-brand illustrations.
• Qwen Edit Plus – an enhanced Qwen-based image editor for precise inpainting, cleanup, and local style changes while preserving overall composition.

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-flex/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