When one company supplies a model and another hosts it for paying users, the economics depend on contract terms most people skim. We draft inference and revenue share agreements that define what counts as revenue, how usage is measured and reported, and what each side may do with the model and its outputs. They work for model developers, inference platforms and the marketplaces between them.
At a glance
- Practice
- AI Infrastructure
- Pricing
- Flat fee, quoted before we start
- Included
- 6 deliverables
- Process
- 3 steps, one named attorney
- Where we work
- Slack, email or a call
When founders call us
The moments that usually bring a company to us for this work.
- 01
An inference platform wants to host your model and pay you a share of usage revenue.
- 02
You run an inference service and are onboarding third-party models under revenue share terms.
- 03
You are listing your model on a marketplace and need to understand its payout and audit terms.
- 04
A partner has stopped reporting usage clearly and you are not sure how your share is calculated.
What you get
Scoped in writing before we start. One flat fee for all of it.
- Revenue share agreement defining gross and net revenue, deductions and payout timing
- Usage metering, reporting and audit rights
- Licence to host and serve the model, with limits on fine-tuning, distillation and redistribution
- Allocation of output rights, acceptable use enforcement and responsibility for end users
- Pricing authority, discounting and minimum guarantees where agreed
- Term, termination and model takedown mechanics
How it works
- 1
Model the economics
We map how money and usage flow so the definitions match reality.
- 2
Draft the rights
We set what the host may do with the model and what the developer keeps.
- 3
Build in verification
We add reporting and audit terms so each side can check the numbers.
Mistakes we help you avoid
The ones we see most often, and catch before they cost you.
- Avoided
Defining revenue as 'net' without listing which deductions are allowed.
- Avoided
No audit right, leaving the model developer to trust the host's reports.
- Avoided
Hosting licences that do not prohibit distilling or fine-tuning the hosted model.
- Avoided
Unclear responsibility for end-user misuse and acceptable use enforcement.
Inference & Revenue Share: questions founders ask
A model developer licenses a model to a host, which serves it to users and pays the developer a share of related revenue. The agreement defines the revenue base, deductions, reporting and payment schedule. It also governs what the host can do with the model.
Common deductions include payment processing fees, taxes, refunds and sometimes compute costs. Each deduction should be listed specifically. Broad catch-all deductions are a frequent source of disputes.
Not unless the agreement allows it, but you should say so explicitly. Many developers prohibit fine-tuning, distillation and training on outputs to protect the model's value. If some modification is allowed, the agreement should address who owns the result.
Usually the party with the end-user relationship enforces acceptable use, but contracts often split responsibilities and indemnities. The right split depends on who controls access, pricing and user terms.
Get a flat-fee quote for Inference & Revenue Share.
A short call with an attorney. You leave with a written scope and one fixed price.
Book a call