bfl/flux-2/klein/4b/base/edit/lora
Flux 2 Klein 4b by BFL - AI-powered image editing, style transfer, and transformation. Edit photos with natural language instructions, remove backgrounds, change styles, and enhance images effortlessly.
PNG, JPEG, WebP, or GIF · 20 MiB maximum each
Example output — click Run to generate your own
Send your first request
OpenAI-compatible endpoint with unified authentication and usage tracking.
https://api.sandbase.ai/v1/runbfl/flux-2/klein/4b/base/edit/loraInput Schema
8 parameters · 1 required · 7 optional
| Parameter | Type | Required | Description |
|---|---|---|---|
prompt | string | Required | The prompt to edit the image. |
seed | integer | Optional | The seed to use for the generation. If not provided, a random seed will be used. |
loras | object[] | Optional | List of LoRA weights to apply (maximum 3). · Default: [] |
images | string[] | Optional | The URLs of the images for editing. A maximum of 4 images are allowed. |
aspect_ratio | string | Optional | The aspect ratio of the generated image. · Options: 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16 21:916:93:24:35:41:14:53:42:39:16 |
output_format | string | Optional | The format of the generated image. · Options: jpeg, png · Default: "png" jpegpng |
guidance_scale | number | Optional | Guidance scale for classifier-free guidance. · Min: 0 · Max: 20 · Default: 5 |
num_inference_steps | integer | Optional | The number of inference steps to perform. · Min: 4 · Max: 50 · Default: 28 |
Output Schema
| Field | Type | Description |
|---|---|---|
id | string | Unique identifier for the generation task |
status | string | Task status: pending, running, completed, failed, timeout |
model | string | Model used for the generation |
outputs | array | Array of output items |
outputs[].url | string | URL of the generated artifact |
outputs[].content_type | string | MIME type (e.g. image/png, video/mp4) |
error | object | null | Error details if failed, null on success |
error.type | string | Machine-readable error type code |
error.message | string | Human-readable error description |
Async Workflow
This model uses asynchronous execution. Submit a request and poll for the result.
- Submit — POST to /v1/run, receive an
id - Poll — GET /v1/run/{id} until status is
completed,failed, ortimeout - Retrieve — Read
outputsfrom the completed response
Code Examples
Ready-to-run snippets
# Step 1: Submit
curl -X POST https://api.sandbase.ai/v1/run \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "bfl/flux-2/klein/4b/base/edit/lora",
"loras": [],
"images": [
"https://static.sandbase.ai/examples/bfl/flux-2/klein/4b/base/edit/lora/input_images_0.png"
],
"prompt": "Change his clothes to casual suit and tie",
"output_format": "png",
"guidance_scale": 5,
"num_inference_steps": 28
}'
# Step 2: Poll result (replace <id>)
curl https://api.sandbase.ai/v1/run/<id> \
-H "Authorization: Bearer YOUR_API_KEY"API README
FLUX.2 [klein] 4B Base LoRA
FLUX.2 [klein] 4B Base LoRA exposes the FLUX.2 Klein 4B Base Edit LoRA capability for reference-led image revision. The undistilled Base architecture retains full training signal and higher output diversity for adaptation and research. A lightweight adapter adds a learned style, subject, or domain concept to this exact path.
Use this 4B route when accessibility, local efficiency, and rapid iteration matter. Describe the requested change separately from source identity, geometry, and regions that must remain fixed. Choose the Base form for fine-tuning or diverse research outputs rather than minimum inference latency.
Highlights
- FLUX.2 Klein 4B Base Edit LoRA architecture. Uses the undistilled 4B foundation with full training signal and diverse output behavior.
- Unified contextual editing. Understands source content and applies directed semantic or appearance changes.
- Adapter-based specialization. Expresses a learned style, character, object, or domain through a lightweight adaptation.
- Reference consistency. Preserves important subject, composition, and detail through the requested revision.
Pricing
| Configuration | Billing unit | Price |
|---|---|---|
| Base generation | Per request | $0.032 |
When to Use
| ✅ Good fit | ❌ Consider alternatives |
|---|---|
| The project needs this exact generation or editing route | The intended task belongs to video, audio, or 3D media |
| Available references match every required local field | Required source assets or rights are unavailable |
| The brief can specify composition, color, and finish | Output must be deterministic at pixel level |
| Supported dimensions and formats fit final placement | Delivery requires unsupported canvas specifications |
| An asynchronous generated image fits production | A live frame-synchronous response is mandatory |
Prompt Guide
Write a complete visual brief: identify the subject and references, define composition and viewpoint, describe lighting, palette, materials, typography, and finish, then list exact preservation constraints. FLUX.2 Klein does not rely on prompt upsampling, so use clear descriptive sentences rather than sparse keywords.
{
"prompt": "A photorealistic editorial composition with precise subject placement, controlled lighting, exact colors, and clean typography",
"seed": 1,
"loras": [],
"images": [
"https://example.com/reference.jpg"
]
}
Technical Specs
| Spec | Value |
|---|---|
| Model ID | bfl/flux-2/klein/4b/base/edit/lora |
| Input fields | seed (integer)<br>loras (array)<br>images (array)<br>prompt (string)<br>aspect_ratio (string; 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16)<br>output_format (string; jpeg, png)<br>guidance_scale (number; 0–20)<br>num_inference_steps (integer; 4–50) |
| Required input | prompt |
| Output fields | url, content_type |
| Execution | Asynchronous job |
Related Models
bfl/flux-2/klein/4bbfl/flux-2/klein/4b/basebfl/flux-2/klein/4b/base/edit