FOR DEVELOPERS

Your LORAs earn stars on others.Here they earn 85%.

List the models you’ve already trained. Set your price. Keep your portfolio anywhere it lives. We take 15% — nothing else.

Your models become licensable AI Cast and Sets in the marketplace — priced by you, cleared for commercial use by buyers.

Request Early Access
  • Your weights. Your IP.
  • Any base model, any rank
  • Pull anytime. Walk with everything.

run-2401 · 8 variants · sdxl

  • rank=8 · lr=1e-4

  • rank=16 · lr=1e-4

  • rank=32 · lr=8e-5

  • rank=64 · lr=5e-5

  • rank=8 · lr=1e-4

  • rank=16 · lr=8e-5

  • rank=32 · lr=5e-5

  • rank=64 · lr=3e-5

$ lookhouse train --base sdxl --variants 8 # 85% to you · 15% to us

  • LIST

    Mirror your Hugging Face, or GitHub LORAs in minutes. Any base model. Any rank.

  • BUILD

    No GPU. No cloud bill. Ship a dataset — get back trained variants.

  • LICENSE

    Free or paid. You pick per LORA. We take 15%, flat.

BRING WHAT YOU BUILT

Your Hugging Face page is your portfolio. Make it a paycheck.

You’ve trained LORAs that get downloaded 10,000 times and earn you nothing. That’s a portfolio. We turn it into inventory.

  • Self-serve uploads via the creator console. You control filenames, trigger words, sample images, and license per LORA.
  • Any base model— SDXL, Flux, SD 1.5, Pony, custom checkpoints. If you can upload it, we host it.
  • Stay listedwherever you want for free and standard licenses. Exclusive options are coming — early uploaders shape the terms.
Request Early Access

Character Cast must have documented performer consent. Style, prop, and location LoRAs upload freely.

import.sources

  • huggingface.co
  • github.com

DEST

lookhouse / marketplace

revenue.share = 0.85

BUILD ENGINE

One dataset in. A library out.

Skip the H100 envy. Train on our engine, keep the weights, keep 85%.

  1. Upload your dataset.

    Captions, tags, reference images — whatever you've got.

    Dataset

    142 images · 38 captions

  2. Configure your run.

    Pick the base model, rank, learning rate, step count, and how many variants you want. We run every config you ask for, in parallel.

    Run config

    SDXL · ranks 16/32/64 · ×6

  3. Get back variants.

    Multiple trained LORAs from one training run — style strengths, ranks, recipes.

    Variants

    4 of 6 complete

  4. List the keepers.

    Test in our preview. Publish the ones that hit. Set the price.

    Listing

    Published · $12

  • You set the spread. We run the GPUs.
  • $0 to list
  • 85% to you
  • 15% on sales

LICENSING

You own it. You price it. You can pull it.

Pick per LORA. Mix tiers across your library. We never touch the floor.

Free / Open Source

$0Listed at no charge.

For the open-source crowd.

Price
$0
Setup
No listing fee
Payout
—
Recommended for paid releases

Standard

85%to you, on every sale.

Set your price per download.

Price
You set it
Setup
No listing fee
Payout
85% / 15%
  • We don't train on your weights.
  • Pull a LORA off the marketplace anytime — buyers keep what they already bought, you stop new sales.
  • Payouts via Stripe, weekly. No minimum. Cash out a dollar if you want.

FAQ

Questions devs actually ask

STOP GIVING IT AWAY

Stop giving it away.

You trained the model. You earned the downloads. Now earn the revenue.

Request Early Access

Verify it’s you

Confirm with a second factor before continuing.

PRIMARY

Use your passkey

Touch ID · Face ID · platform authenticator

Tap below to prompt your device

Or use a code instead

We’ll skip this verification for the next 5 minutes on this device.