---

name: black-forest-labs-flux-dev-lora description: "Black Forest Labs | Flux Dev Lora. Generate images with Flux Dev and LoRA weights from multiple sources. Triggers: text to image, flux dev, lora, bfl, image generation, civitai, huggingface" allowed-tools: Bash(curl *), WebFetch

Black Forest Labs | Flux Dev Lora

Generate images with Flux Dev model and custom LoRA weights from Replicate, HuggingFace, CivitAI, or direct URLs. Supports image-to-image mode and multiple output formats.

Quick Start

Requires an each::labs API key. Get one at eachlabs.ai.

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "black-forest-labs-flux-dev-lora",
    "version": "0.0.1",
    "input": {
      "prompt": "a serene lake at twilight with reflections, photorealistic",
      "lora_weights": "https://example.com/nature-lora.safetensors",
      "lora_scale": 0.8,
      "aspect_ratio": "16:9",
      "guidance": 3
    }
  }'

Parameters

Parameter	Type	Default	Description
aspect_ratio	string	1:1	Aspect ratio of the output. enum: 1:1, 16:9, 21:9, 3:2, 2:3, 4:5, 5:4, 3:4, 4:3, 9:16, 9:21
civitai_api_token	string		CivitAI API token for authenticated LoRA downloads.
disable_safety_checker	boolean	false	Disable safety checker for generated images.
extra_lora	string		Additional LoRA weights URL. Supports Replicate and HuggingFace formats.
extra_lora_scale	number	1	Scale for the extra LoRA (0-2).
go_fast	boolean	false	Run with fp8 quantized model for faster predictions.
guidance	number	3	Guidance scale for generated image.
hf_api_token	string		HuggingFace API token for authenticated LoRA downloads.
image	string		Input image for image-to-image mode. Output matches input aspect ratio.
lora_scale	number	1	Scale for the main LoRA (0-2).
lora_weights	string		Main LoRA weights. Supports Replicate, HuggingFace, CivitAI, or URL.
megapixels	string	1	Output megapixels. enum: 1, 0.25
num_inference_steps	integer	28	Number of denoising steps (28-50 recommended).
num_outputs	integer	1	Number of outputs to generate.
output_format	string	webp	Output format. enum: webp, jpg, png
output_quality	integer	80	Quality for output images (0-100).
prompt	string		Prompt for generated image.
prompt_strength	number	0.8	Prompt strength for img2img mode (1.0 = full destruction).
seed	integer		Random seed for reproducible generation.

Examples

LoRA from HuggingFace:

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "black-forest-labs-flux-dev-lora",
    "version": "0.0.1",
    "input": {
      "prompt": "anime girl in a cherry blossom garden, detailed",
      "lora_weights": "username/anime-style-lora",
      "lora_scale": 0.9,
      "aspect_ratio": "3:4",
      "num_inference_steps": 30,
      "output_format": "png"
    }
  }'

Image-to-image with dual LoRAs:

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "black-forest-labs-flux-dev-lora",
    "version": "0.0.1",
    "input": {
      "prompt": "vintage film photography style, warm tones",
      "image": "https://example.com/input-photo.jpg",
      "lora_weights": "https://example.com/vintage-lora.safetensors",
      "lora_scale": 0.7,
      "extra_lora": "https://example.com/film-grain-lora.safetensors",
      "extra_lora_scale": 0.5,
      "prompt_strength": 0.6,
      "seed": 42
    }
  }'

Related Models

• BFL Flux Lora - BFL Flux with LoRA support
• Flux 2 | Text to Image Lora - Flux 2 with LoRA support

Documentation

• each::labs Docs
• API Reference