Publishing Models

Any organization on JAOT can publish a model to the marketplace, making it available for other teams to discover and fork into their own studios. Publishing lets you share your domain expertise with users who need optimization but may not have the skills to formulate it. The marketplace is free and collaborative — there are no prices or commissions.

How Publishing Works

Publishing attaches a public listing to your model project. The listing pins your latest committed version — the marketplace never shows your work-in-progress draft. When another team clicks Use in studio, they fork that pinned version into a model of their own; your project stays untouched and keeps evolving independently.

Who Can Publish

Publishing is available to any authenticated user for the model projects of their organization. The model must have at least one committed version — the listing needs a stable snapshot to pin.

Preparing Your Model

Before publishing, make sure your model:

Solves correctly -- Run it from the Solve tab and check the result
Is committed -- Commit the state you want to share, with a meaningful "What changed?" message
Is understandable -- A clear name, sensible variable/constraint names, and a description of what it optimizes

Publishing Process

Step 1: Commit a version

In the Model Workspace, commit the draft you want to share (Versioning). Publishing pins this committed version.

Step 2: Open the publish page

Go to the Analyze tab and find the Marketplace card, then click Publish to marketplace. This opens the publish form at /studio/{model}/publish.

Step 3: Fill in the Publishing Form

Field	Required	Description
Display name	Yes	The public-facing name shown in the marketplace (e.g., "Warehouse Location Optimizer")
Description	Yes	A detailed explanation of what the model solves and how it works
Short description	No	A one-line summary for the model card in search results
Category	Yes	The industry category (manufacturing, finance, logistics, etc.)
Tags	No	Keywords to improve discoverability (e.g., "routing", "scheduling", "mixed-integer")
Overview section	No	Extended description with use cases and benefits
Features section	No	Key capabilities and model characteristics
How it works	No	Walkthrough of the model's optimization approach
Example I/O	No	Sample input and output for the documentation
Changelog	No	Version history and changes

Step 4: Publish

Click Publish. Your model appears in the public catalog immediately — there is no review queue. You are redirected to its public marketplace page.

Managing a Published Model

Reopen the publish page from the same Marketplace card on the Analyze tab. In edit mode you can:

Update the presentation -- name, descriptions, category, tags, and sections take effect immediately
Add media -- upload a logo and screenshots once the listing exists
Re-pin a newer version -- commit again and publish to move the listing to your latest committed version. Teams that already forked keep their copies; new forks get the new version

Free and Collaborative

The marketplace is free to use — models carry no price and no commission is charged. Publishing builds your reputation through adoption, ratings, and reviews.

Tip: A clear description, a working example, and accurate categorization drive adoption far more than anything else.

Adoption Metrics

Every listing tracks its adoption, visible on its marketplace page and on your public author profile:

Metric	Description
Adoptions	Total number of teams that forked your model into their studio
Executions	Total number of times the model has been solved
Rating	Average user rating and individual reviews

You also get an in-app notification each time a team adopts one of your models (configurable in your workspace settings). Use these signals to understand demand, identify opportunities for new models, and improve your published ones.

Publishing via API

You can also publish programmatically — the endpoint takes a project with at least one committed version:

import httpx
 
headers = {"Authorization": "Bearer ok_live_your_key_here"}
 
response = httpx.post(
    "https://jaot.io/api/v2/projects/mp_abc123/publish",
    headers=headers,
    json={
        "display_name": "Warehouse Location Optimizer",
        "description": "Finds optimal warehouse locations to minimize total shipping cost...",
        "short_description": "Optimize warehouse placement for minimum logistics cost",
        "category": "logistics",
        "tags": ["warehouse", "facility-location", "logistics"],
    },
)
 
published = response.json()
print(f"Published: {published['display_name']} (listing ID: {published['id']})")

A project without a committed version returns 400. See the Models API reference for the full endpoint specification.

Best Practices

Write a thorough description. Users decide whether to fork based on the detail page. Explain the problem, the approach, expected inputs, and what kind of results to expect.
Commit deliberately. The listing pins a committed version — make it a state you are proud of, with a clean history behind it.
Choose the right category. Accurate categorization helps users find your model through filters and improves marketplace discovery.
Respond to reviews. Engage with users who leave feedback. Address issues promptly and update your model based on common requests.
Keep the changelog updated. When you re-pin a new version, document the changes so adopters know what improved.