Flux Dev LoRA API Documentation

Flux Dev LoRA API Documentation

Playground

FLUX.1 [dev] endpoint with LoRA support for fast, high-quality image generation and simple personalization via pre-trained LoRA adapters. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

FLUX.1 [dev] is a 12B parameter rectified flow transformer for advanced text-to-image generation. It supports prompt-only generation as well as image inpainting and LoRA customization, making it a flexible tool for both research and creative workflows.

Why it looks great

• High-quality output: Cutting-edge visual fidelity, second only to FLUX.1 [pro].
• Prompt alignment: Strong competitive prompt following, rivaling closed-source alternatives.
• Efficient training: Trained with guidance distillation for better speed-performance balance.
• Flexible editing: Supports image + mask editing, LoRA fine-tuning, and custom strength control.
• Open weights: Enables research, experimentation, and innovative creative pipelines.

Limits and Performance

• Max resolution: up to 1536 × 1536 pixels

• Optional inputs:

• image (for img2img)

• mask_image (for inpainting)

• LoRA support: add multiple .safetensors with adjustable scale

• Inference controls:

• num_inference_steps (default ~28)

• guidance_scale (default ~3.5)

• strength (the strength of transform the reference image)

• Output format: JPEG / PNG / WEBP

• Seed: reproducibility (-1 = random)

Pricing

Just $0.015 per image !!

How to Use

1. Write a prompt — detailed scene + style (lighting, realism, mood).
2. (Optional) Upload an image to guide generation.
3. (Optional) Add a mask image for inpainting.
4. Adjust parameters:

• Strength (the strength of transform the reference image).
• LoRAs (add path/URL + scale).
• Size (width & height, up to 1024×1024).
• Inference steps and guidance scale.

5. Set num_images (default 1).
6. (Optional) Fix seed for reproducibility.
7. Choose output format and run.

Pro tips

• Use higher inference steps for more detail, lower for speed.
• Adjust guidance scale to balance prompt strength vs. creativity (3–7 recommended).
• Apply mask + strength for clean local edits (inpainting).
• Blend multiple LoRAs for hybrid style outputs.
• Use consistent seeds when testing parameter changes for controlled comparison.

Notes

• The image URL must be valid and accessible; otherwise, the job may fail.
• For mask_image, do not upload the original or unprocessed image directly — ensure the mask is correctly prepared.
• LoRA files must be uploaded from trusted platforms and set to public access to be usable.
• Parameters such as num_inference_steps (and others) directly affect runtime: the larger the value, the longer the generation will take.

Reference

• Use your LoRA

• Train your LoRA

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",
  "strength": 0.8,
  "size": "1024*1024",
  "num_inference_steps": 28,
  "guidance_scale": 3.5,
  "num_images": 1,
  "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/flux-dev-lora" \
  -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