Shopify
WooCommerceVERTEX AI TRY-ON ALTERNATIVE
Genlook vs. Google Vertex AI Virtual Try-On
Google's Virtual Try-On model renders clothing very well, at $0.06 an image inside a Google Cloud project. Genlook is a try-on API at $0.02 that covers every product type and handles the work around the model.
Written by Thibault Mathian, founder of Genlook. Genlook makes one of the two products compared here. Competitor facts were last checked on October 7, 2026 against their public listings and docs, linked in Sources.
01 · The short verdict
A strong model, or a finished try-on API.
Both produce excellent clothing try-ons. They differ in price, product coverage and how much you build around the call.
Google Vertex AI
Built for Google Cloud teams
- Excellent clothing results in our benchmark, close to Genlook's
- Runs in your Google Cloud project, EU regions included
- Up to 4 output images per request
- $0.06 per image; $0.048 only on a 3-year savings plan
- Clothing only, one product per request, synchronous calls
- virtual-try-on-001 is deprecated; retirement set for March 2027
Genlook
Built for shipping try-on
- $0.02 per try-on on plans, down to $0.015
- Any product type: clothing, shoes, glasses, jewelry, hats, wigs
- Async with webhooks, plus batches at half a credit
- One stable contract while the engine improves behind it
02 · Feature by feature
Where each API stands.
Checked against Google Cloud's model pages, pricing and release notes on the date above.
Price per try-on
$0.02 on plans, down to $0.015; $0.04 pay as you go
$0.06; $0.054 or $0.048 on 1- or 3-year savings plans
Free start
10 free credits, keys in minutes
$300 Google Cloud credit for new customers
Speed
8.2s median
No published figure; 11 to 12.5s in our benchmark
Product types
Any product type, one endpoint
Clothing; no published category list
Delivery
Sync call, or async with webhooks; batches at half a credit
Synchronous predict call; no batch inference
Model lifecycle
Same endpoint; engine updates behind it
Deprecated Sept 14, 2026; retirement March 15, 2027
Setup
An API key
Google Cloud project, billing, IAM and the Vertex AI API
End-user data
User records, deletion API, image auto-expiry
Not offered; storage and records are yours
Data use
Never used to train models
Not used for training without your permission
Same photos, same day
Genlook and Vertex AI, side by side.
From our September 2026 benchmark: the same shopper photo and product photo sent to each API, default settings, one run each, unretouched. Times and prices are what each run took and cost; Genlook's time is its production median at the time.
See all 10 APIs on 8 garmentsTrench coat, packshot
Vertex dressed the coat but removed the shopper's jeans. Genlook kept them and matched the knee length.
Lace wedding gown
Both render the lace bodice and tulle skirt faithfully. A close call.
Football shirt with sponsor text
Both keep the sponsor text readable and the stripes aligned.
03 · The real difference
A model endpoint, or the whole try-on layer
On clothing, Google's model is one of the best we have tested. In our September benchmark it matched Genlook on most garments, and its only unwanted change in eight tries was removing a shopper's jeans under a long coat. If your stack already lives in Google Cloud and you only need garments, it is a serious option.
What you get is a model endpoint. Each call takes one person image and one product image and returns up to four images synchronously; there are no webhooks, no batch mode and no records of your end users. Coverage is clothing, with no published category list for shoes, glasses or jewelry. And the model has its own lifecycle: virtual-try-on-001 reached general availability in January 2026, was marked deprecated on September 14, 2026, and is set to retire in March 2027, with Google's docs now listing Gemini image models as try-on options. Each change is a migration on your side.
How Genlook handles it
Genlook sells the try-on layer as one API. The same endpoint covers clothing, shoes, glasses, jewelry, hats and wigs, answers synchronously or through webhooks, and runs large jobs as batches at half a credit. End users are records you can delete on request, uploaded images expire on a schedule you choose, and when a better model appears the engine changes behind the same contract, listed in the changelog. A try-on costs $0.02 on a plan and $0.015 at volume, a third of Vertex's list price.
If you need clothing only, run everything inside Google Cloud and want to own the integration, Vertex is a strong pick. If you want every product type, a stable contract and the operations included, start with Genlook.
04 · In practice
What the Genlook API is tuned for.
Any product type
Clothing, shoes, glasses, jewelry, hats and wigs through one endpoint, with no category parameter to manage.
A contract that stays put
Engine improvements land behind the same endpoint. No model ids to track, no deprecation notices to plan around.
Async when you need it
Webhooks for completed and failed generations, and batches of up to 500 try-ons at half a credit each.
End-user data handled
User records, a deletion endpoint for privacy requests, and image retention of 1, 3 or 7 days.
05 · Switching
Moving from Vertex AI.
Create a key
Self-serve at platform.genlook.app, 10 free credits included. No cloud project, billing account or IAM setup.
Replace the predict call
Send the same person and product images to Genlook's try-on endpoint, synchronously or with a webhook. Shoes, glasses and jewelry work through the same call.
Go live
Plans start at $20 a month for 1,000 try-ons ($0.02 each), down to $0.015 at volume, or pay as you go at $0.04. No model deprecation dates to track.
06 · FAQ
Questions, answered.
How much does Google's Virtual Try-On cost?↓
$0.06 per generated image on Vertex AI. Google's savings plans bring it to $0.054 with a 1-year commitment or $0.048 with a 3-year commitment. Genlook costs $0.02 per try-on on a plan, down to $0.015 at volume, or $0.04 pay as you go.
Is Vertex AI's try-on better than Genlook's?↓
On clothing, the two were close in our benchmark: both rendered lace, prints and sponsor text faithfully. Vertex removed a shopper's jeans under a long coat once; Genlook made no unwanted changes in that run. The bigger differences are coverage (Vertex is clothing only), price and the integration work around the call.
Is virtual-try-on-001 being retired?↓
Google's model page lists a deprecation date of September 14, 2026 and a retirement date of March 15, 2027, and its try-on guide now also lists Gemini image models. Plan a migration if you build on it. Genlook keeps one endpoint and updates the engine behind it.
Does Vertex AI try-on work for shoes, glasses or jewelry?↓
Google's docs describe it for clothing and publish no category list for accessories. Genlook covers clothing, shoes, glasses, jewelry, hats and wigs through the same endpoint.
Can I test both?↓
Yes. New Google Cloud customers get $300 in credit, and Genlook accounts start with 10 free credits. Send the same photos to both and compare the results on your own catalog.
Sources
- virtual-try-on-001 model page: launch stage, deprecation and retirement dates, specs, regions
- Generate virtual try-on images: supported models, REST and Python usage, free credits
- Virtual try-on API parameters: sample count, safety settings, watermark
- Vertex AI generative AI pricing: $0.06 per image, savings plan prices
- Vertex AI release notes: preview, quality updates, general availability
- Vertex AI virtual-try-on-001 model page: request quota
- Genlook Try-On API pricing
Checked on October 7, 2026. Plans and features change; if something here is out of date, write to hello@genlook.app and we will correct it.
Build try-on into your product.
Self-serve keys, ten free credits, and two API calls to your first generation.