You're looking at a specific version of this model. Jump to the model overview.
replicate /flux-3-klein-image-lora:522cd456
Input schema
The fields you can use to run this model with an API. If you don’t give a value for a field its default value will be used.
| Field | Type | Default value | Description |
|---|---|---|---|
| prompt |
string
|
A full scene caption describing the requested image.
|
|
| distilled |
boolean
|
True
|
Enable the extracted distilled adapter and 16-step flow-map sampling. Disable for the 50-step CFG base model.
|
| width |
integer
|
1024
Min: 64 Max: 4096 |
None
|
| height |
integer
|
1024
Min: 64 Max: 4096 |
None
|
| num_inference_steps |
integer
|
0
Max: 100 |
0 uses the selected model's default: 16 distilled or 50 base.
|
| guidance |
number
|
3.5
Min: 1 Max: 10 |
CFG scale for the base model. Ignored in distilled mode.
|
| seed |
integer
|
-1
|
Random seed. -1 chooses a random seed.
|
| upsample_prompt |
boolean
|
True
|
Expand the prompt with scene, composition, and lighting details before generating. Disable to use your prompt exactly as written. This rewrites text; it does not upscale the image.
|
| lora_url |
string
|
Optional direct HTTP(S) download URL for a FLUX.3 Klein LoRA safetensors file. Supports PEFT/SimpleTuner and ComfyUI LoRAs up to rank 128 per projection. Include the LoRA's trigger in your prompt; disable prompt upsampling to preserve it exactly.
|
|
| lora_scale |
number
|
1
Max: 2 |
Strength of the LoRA supplied in lora_url. 0 disables it; 1 uses its normal strength. Independent of the distilled toggle.
|
Output schema
The shape of the response you’ll get when you run this model with an API.
Schema
{'format': 'uri', 'title': 'Output', 'type': 'string'}