LongCat Image Edit: AI Image Editor + Text Rendering

Docs Blog Pricing Enterprise Careers Research Grants

fal-ai/longcat-image/edit

LongCat image Edit is a 6B parameter image editing model excelling at multilingual text rendering, photorealism and deployment efficiency.

Inference

Commercial use

Schema

LLMs

Playground API

Input

Prompt*

Image URL*

Hint: Drag and drop image files from your computer, images from web pages, paste from clipboard (Ctrl/Cmd+V), or provide a URL. Accepted file types: jpg, jpeg, png, webp, gif, avif

Additional Settings

Customize your input with more control.

Result

Idle

What would you like to do next?

Download

{
  "images": [
    {
      "height": 768,
      "content_type": "image/png",
      "url": "https://storage.googleapis.com/falserverless/example_outputs/longcat_image/edit.png",
      "width": 1024
    }
  ],
  "prompt": ""
}

Your request will cost $0.15 per megapixel.

Logs

Longcat Image | [image-to-image]

Longcat Image Edit is a 6B parameter image editing model delivering natural language transformations at $0.15 per megapixel. Trading raw generation speed for context-aware editing precision, it handles multilingual text rendering and photorealistic modifications without requiring masks or layers. Built for developers who need semantic understanding in their editing pipeline, not just pixel manipulation.

Built for: Natural Language Photo Edits | Multilingual Text Overlays | Context-Aware Image Transformations

Performance That Scales

At $0.15 per megapixel, Longcat Image positions between budget batch processors and premium editing APIs, prioritizing semantic understanding over raw throughput.

Metric	Result	Context
Parameter Count	6B	Mid-size architecture balancing quality and inference cost
Inference Steps	1-50 (default 28)	Configurable quality/speed tradeoff via API parameters
Cost per Edit	$0.15/megapixel	Scales with output resolution, not complexity
Batch Capacity	1-4 images	Parallel processing for workflow efficiency
Acceleration Modes	none/regular/high	Hardware optimization tiers for latency-sensitive applications

Instruction-First Editing Without Masking

Longcat Image processes natural language editing commands directly against reference images, eliminating the traditional mask-layer-apply workflow. The 6B parameter architecture interprets spatial relationships and contextual intent, where "add elegant cursive text with lightning streaks at the top" executes as a single atomic operation.

What this means for you:

Multilingual text rendering: Handles non-Latin scripts and complex typography without font management overhead
Photorealistic integration: Edits respect depth, perspective, and lighting from the source image automatically
Flexible acceleration: Three hardware tiers (none/regular/high) let you trade latency for cost based on workload priority
Output format control: JPEG/PNG/WebP export with configurable safety filtering for production pipelines

Technical Specifications

Spec	Details
Architecture	Longcat Image Edit 6B
Input Formats	PNG, JPEG, WebP, GIF, AVIF via URL
Output Formats	JPEG, PNG, WebP
Resolution Range	Variable (priced per megapixel)
License	Commercial use permitted

API Documentation | Quickstart Guide | Pricing

How It Stacks Up

FASHN Virtual Try-On V1.5 Image to Image – Longcat Image prioritizes general-purpose editing through natural language commands at $0.15/megapixel. FASHN specializes in garment-specific transformations with body-aware fitting for fashion and e-commerce workflows where clothing visualization accuracy justifies specialized architecture.

Longcat Image Image to Image