All models
Images

GFPGAN Restore

Restore and sharpen faces and old or degraded photos, through one API.

Modality
Text → Image
Capacity
Up to 2K resolution
Output
PNG · JPEG · WebP
Latency
Seconds per image
$0.01$0.012520% off
Category
Images
Provider
fal.ai
Model id
gfpgan
Billing
Pay-as-you-go
Service advantages
  • 99.9% SLA
  • Official discount
  • Pay-as-you-go
  • Fast delivery

Try it

Type a prompt and preview a sample response. Real runs use your API key — pay only on success.

Input
Try an example
Aspect ratio
Seed

Est. cost: $0.010 / run

Output
Your result will appear here
Sign in to run for realSign in
Request
curl https://api.modelplex.ai/v1/images/generations \
  -H "Authorization: Bearer $MODELPLEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gfpgan","prompt":"A studio product shot of a ceramic cup on linen, soft morning light","size":"1024x1024","seed":204815}'

Overview

GFPGAN Restore is fal.ai's image-generation model, available through Modelplex. Restore and sharpen faces and old or degraded photos, through one API. Call it with the OpenAI-compatible API — one key, one balance, transparent pricing, and you only pay for what succeeds.

50K+
active developers
99.9%
uptime
faster routing
70%
avg savings

Pricing details

Unit price
$0.01

per image

Original price
$0.0125
Discount
−20%
Billing
Pay-as-you-go

Prices include the official discount and you're charged only on success — failed generations are free.

Why it stands out

What you can build

E-commerce & product shotsMarketing key visuals & adsUI / UX mockupsSocial media graphicsStoryboards & concept artLocalized creative at scale

Up and running in 3 steps

  1. 1

    Create an account

    Sign up and grab free credits — no card required.

  2. 2

    Point the SDK

    Set the base URL and your APIMart key.

  3. 3

    Make a call

    Use the model id and pay only on success.

FAQ

Point the OpenAI SDK at Modelplex, set your key, and use the model id "gfpgan". The base URL and key are the only change to your code.

$0.01, billed per call with our standard discount applied up front — and you're only charged on success.

Yes — new accounts start with free credits, so you can try GFPGAN Restore before adding a balance.

Served on a 99.9% uptime target with automatic same-kind fallback if an upstream has a hiccup.

Yes. It's drop-in OpenAI-compatible — no code changes beyond the base URL and key.

Square, portrait and landscape from 512×512 up to 2K, selectable per request via the size parameter.

Where GFPGAN Restore supports it, pass a reference image to guide style and composition, or a mask to edit specific regions.

As image URLs (PNG/JPEG/WebP) re-hosted on Modelplex and served through short-lived signed links.

Per-key rate limits you set from the console, with sensible defaults on every new key. Need more headroom? Just ask support.

Calls are served from regional infrastructure on a 99.9% uptime target, and your inputs and outputs are never used to train models.

Related models