Licensing catalog

The licensing reference for AI Inference: what can be licensed, who licenses to whom, and how governance and economics are divided. Pair it with the commercial catalog.

What can be licensed

AI Inference is a managed inference routing and metering platform. It accepts an inference request from any application, resolves it to a model in its catalog — the one the request names, or the lowest-cost enabled model offering the capability requested — dispatches the call, counts the exact input and output tokens, and prices them at the rate published for that model. One metered usage event is recorded per request, under an idempotency key, so a retry cannot be billed twice.

The platform is licensable to an enterprise that wants to operate its own branded inference surface without building the routing, metering and billing layer itself.

Who can license to whom

• Ventureship licenses the AI Inference platform to a licensee — an enterprise, a platform or a sovereign.
• The licensee operates the platform under its own brand, within its defined territories and verticals.
• Ventureship retains governance of the platform itself and of the model catalog, including which entries exist, how they are described to end users, and the price published against each one.
• End users interact only with the licensee's branded surface.

Governance

The licensing agreement grants Ventureship ownership and control of the platform across use, future direction, marketing and sales, accounting and finance, legal, insurance and risk management, employment, production, operations and fulfilment, and promotion, through the parent company, Ventureship.

The economic share for a licensee is structured as a revenue share on metered usage, with a floor and a cap defined in the operative agreement. The reference agreement is published at mipl.ventureship.xyz, and the operative agreement for a given licence is the one signed by the parties — not this page.

What a licensee operates

• The branded inference surface, its playground and its request history.
• The console: accounts, per-account requests-per-minute limits, per-account cache windows, and the catalog entries enabled for their market.
• Metered usage reporting and the billing relationship with their own end customers.

What is not warranted

The platform is licensed as it is. There is no uptime commitment or service-level agreement, no security certification or audit attestation, and no warranty as to the accuracy, safety or fitness of a model's output. Any assurance a licensee needs beyond that has to be written into the operative agreement.

About this catalog

This is the canonical licensing reference for AI Inference and is intended to be cited as such. The commercial catalog covers the product as a customer buys it.