Construction AI models,
ready to deploy.

The marketplace for YOLO models, training datasets, and LLMs built for blueprints by the people who label them.

How it works

Model ExchangeMarketplaceSellerModels + DataBuyerSearch + DeployUpload + metricsBrowse / DownloadS3Model filesPostgreSQLListings + reviewsStripePayments (soon)Reviews / Custom requestsUpload flowDownload flowP2P messagesStorage

Browse by category

For buyers

  • Verified training metrics (mAP, precision, recall)
  • Sample predictions on real blueprints
  • SHA-256 verified downloads
  • 5 downloads per purchase, 72hr access

For sellers

  • Keep up to 90%+ of every sale
  • Stripe Connect pays you directly
  • Set your own price and license
  • No exclusivity — list anywhere

Only construction

  • CSI division + trade tags
  • Blueprint-specific categories
  • Buyers who need your exact use case
  • No noise from general AI marketplaces

Built to support the BlueprintParser ecosystem

Model Exchange is part of the open-source BlueprintParser community. Labelers, engineers, and contractors collaborate to build blueprint AI — and this marketplace gives that work a place to be shared, sold, and reused across projects.

Why the marketplace exists

  • BlueprintParser is the open-source AI takeoff tool that makes blueprints understandable to LLMs.
  • Model Exchange gives BP users a place to buy and sell blueprint-trained models and datasets.
  • The goal is to turn blueprint AI work into reusable products instead of isolated one-off effort.

Why collaboration matters

  • A serious electrical object detection stack can mean 50 classes or more.
  • That takes thousands of hours of labeling, testing, and retraining to do well.
  • The marketplace lets multiple contributors build that ecosystem together and monetize the work.

Why buyers benefit

  • Teams can budget for targeted models instead of paying for expensive per-seat takeoff software.
  • A strong blueprint-trained model can be cheaper than another $350/month seat.
  • Buy once, deploy in your own BlueprintParser workflows, and improve from there.