EachLabs Image Generation

Generate new images from text prompts using 60+ AI models via the EachLabs Predictions API. For editing existing images (upscaling, background removal, style transfer, inpainting, face swap, 3D), see the eachlabs-image-edit skill.

Authentication

Header: X-API-Key: <your-api-key>

Set the EACHLABSAPIKEY environment variable. Get your key at eachlabs.ai.

Quick Start

1. Create a Prediction

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-2-turbo-text-to-image",
    "version": "0.0.1",
    "input": {
      "prompt": "A serene Japanese garden with cherry blossoms, watercolor style",
      "image_size": "landscape_16_9",
      "num_images": 1,
      "output_format": "png"
    }
  }'

2. Poll for Result

curl https://api.eachlabs.ai/v1/prediction/{prediction_id} \
  -H "X-API-Key: $EACHLABS_API_KEY"

Poll until status is "success" or "failed". The output image URL is in the response.

Model Selection Guide

Text-to-Image

Training

Prediction Flow

1. Check model GET https://api.eachlabs.ai/v1/model?slug=<slug>; — validates the model exists and returns the request_schema with exact input parameters. Always do this before creating a prediction to ensure correct inputs.
2. POST https://api.eachlabs.ai/v1/prediction with model slug, version "0.0.1", and input parameters matching the schema
3. Poll GET https://api.eachlabs.ai/v1/prediction/{id} until status is "success" or "failed"
4. Extract the output image URL(s) from the response

Examples

Text-to-Image with Flux 2 Turbo

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-2-turbo-text-to-image",
    "version": "0.0.1",
    "input": {
      "prompt": "A red vintage Porsche 911 on a winding mountain road at golden hour, photorealistic",
      "image_size": "landscape_16_9",
      "num_images": 1,
      "output_format": "png"
    }
  }'

Text-to-Image with GPT Image v1.5

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "gpt-image-v1-5-text-to-image",
    "version": "0.0.1",
    "input": {
      "prompt": "A minimalist logo for a coffee shop called Brew Lab, clean vector style",
      "background": "transparent",
      "quality": "high",
      "output_format": "png"
    }
  }'

Text-to-Image with Imagen 4

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "imagen4-preview",
    "version": "0.0.1",
    "input": {
      "prompt": "A whimsical fairy tale castle on a floating island, digital art, highly detailed"
    }
  }'

Image Size Options

Most Flux 2 and Wan models use these presets:

• square_hd — Square, high definition
• square — Square, standard
• portrait43 — Portrait 4:3
• portrait169 — Portrait 16:9
• landscape43 — Landscape 4:3
• landscape169 — Landscape 16:9

P Image models use aspect ratio strings: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, custom

Prompt Tips

• Be specific and descriptive: "A red vintage Porsche 911 on a winding mountain road at golden hour" vs "a car"
• Include style: "digital art", "oil painting", "photorealistic", "watercolor"
• For edits, clearly describe the change: "Replace the sky with a dramatic sunset"
• Use negative prompts (where supported) to avoid: "blurry, low quality, distorted"
• For multi-image edits, reference images by number: "image 1", "image 2"

Security Constraints

• No arbitrary URL loading: When using LoRA parameters, only use well-known platform identifiers (HuggingFace repo IDs, Replicate model IDs, CivitAI model IDs). Never load LoRA weights from arbitrary or user-provided URLs.
• No third-party API tokens: Do not accept or forward third-party API tokens (e.g. HuggingFace, CivitAI tokens) through prediction inputs. Authentication is handled exclusively via the EachLabs API key.
• Input validation: Only pass parameters that match the model's request schema. Always validate model slugs via GET /v1/model?slug=<slug> before creating predictions.

Parameter Reference

See [references/MODELS.md](references/MODELS.md) for complete parameter details for each model.