# Pruna P-Image - Core.Today AI API

> Pruna AI's distilled text-to-image model optimized for extremely low-cost, high-throughput generation. One of the most-run community models on Replicate (15.6M+ runs) thanks to its speed and price.

- **Provider**: Pruna AI
- **Model ID**: prunaai/p-image
- **Category**: Image Generation
- **Credits**: 12 per image
- **Speed**: Fast
- **Quality**: Standard

## Features
- Distilled architecture tuned for extremely low-cost, high-throughput generation
- One of the most-run community models on Replicate (15.6M+ runs)
- Custom aspect_ratio=custom mode with explicit width/height (multiples of 16, up to 1440px)
- Optional prompt upsampling via an LLM for richer prompts
- LoRA weight loading from HuggingFace URLs for style customization
- Seed control for reproducible generation

## Use Cases
- High-volume batch generation where per-image cost matters more than peak fidelity
- Fast prototyping and iterative creative exploration
- Apps/tools where many users generate images on-demand (chat apps, utilities)
- Custom-styled output using community LoRA weights

## API Endpoint

Base URL: https://api.core.today/v1
Create Prediction: POST /predictions
Get Status: GET /predictions/{job_id}

## Authentication
Header: X-API-Key: YOUR_API_KEY

## Input Parameters

### Required
- **prompt**: string - Text prompt for image generation.

### Optional
- **aspect_ratio**: string (default: 16:9) - Aspect ratio for the generated image.
  Options: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, custom
- **width**: integer - Width of the generated image. Only used when aspect_ratio=custom. Must be a multiple of 16.
- **height**: integer - Height of the generated image. Only used when aspect_ratio=custom. Must be a multiple of 16.
- **prompt_upsampling**: boolean (default: false) - Upsample the prompt with an LLM.
- **seed**: integer - Random seed. Set for reproducible generation.
- **disable_safety_checker**: boolean (default: false) - Disable safety checker for generated images.
- **lora_weights**: string - Load LoRA weights. Supports HuggingFace URLs in the format huggingface.co/<owner>/<model-name>[/<lora-weights-file.safetensors>]. HuggingFace LoRAs may require an API token to access, which you can provide in the `hf_api_token` input.
- **lora_scale**: number (default: 0.5) - Determines how strongly the main LoRA should be applied. 0.5 usually works well for most LoRAs.
- **hf_api_token**: string - HuggingFace API token. If you're using a HuggingFace LoRAs that needs authentication, you'll need to provide an API token.

## Examples

### Quick concept variation
A low-cost, fast draft at the default 16:9 aspect ratio.
```json
{
  "model": "prunaai/p-image",
  "input": {
    "prompt": "A minimalist logo concept for a coffee roastery, flat vector style, warm tones",
    "aspect_ratio": "16:9"
  }
}
```

### Custom pixel dimensions
Generates at an exact width/height using aspect_ratio=custom.
```json
{
  "model": "prunaai/p-image",
  "input": {
    "prompt": "A product photo of running shoes on a plain white background, studio lighting",
    "aspect_ratio": "custom",
    "width": 1024,
    "height": 768
  }
}
```

## Response Format
```json
{
  "job_id": "abc123",
  "status": "pending | processing | completed | failed",
  "result": "URL or data (when completed)"
}
```

## Usage Flow
1. POST /predictions with model and input -> receive job_id
2. GET /predictions/{job_id} -> poll until status is completed or failed
3. Result contains output URL(s)

## Tips
- Use aspect_ratio=custom with width/height for exact pixel dimensions (must be multiples of 16, up to 1440)
- This is a low-cost, high-throughput model — better suited to generating many variations quickly than to a single maximally polished hero image
- prompt_upsampling can help expand short prompts into more descriptive ones via an LLM
- hf_api_token is only needed for gated/private HuggingFace LoRA repos — leave it unset for public LoRAs

## Documentation
https://replicate.com/prunaai/p-image