Provider Agreement

This Provider Agreement applies to anyone who applies to supply AI API capacity to Onlist as an upstream provider. It supplements the Terms of Service and Privacy Policy.

As an upstream provider, you supply infrastructure and capacity to Onlist; Onlist meters that usage and is the party that delivers the service to, bills, and supports the buyer. You are an upstream supplier to Onlist, not a direct seller to the buyer. This Agreement sets out your obligations to Onlist; it does not make you the buyer-facing service provider.

If you cannot meet these obligations, do not submit a provider application.

1. Provider identity

You must provide accurate business or operator identity, contact, endpoint, model, pricing, and capacity information. You may not impersonate another provider, operate under multiple personas to manipulate ranking, or imply that Onlist endorses your service beyond your participation as an upstream supplier.

You must be at least 18 years old and able to form a binding contract. If you apply on behalf of a business, you confirm that you are authorized to accept this Agreement for it.

2. Upstream declaration

You must accurately disclose how your endpoint reaches the model:

• Direct: you connect to the model provider's official API (for example, Anthropic Messages API, OpenAI Chat Completions) using credentials you are legally entitled to use.
• Intermediated: you reach the model through additional third-party infrastructure that sits in front of the model provider. You must disclose this, and you remain responsible for what that third party does with the data.
• Self-hosted: you run the model weights on infrastructure you operate or rent.

Misrepresenting how you reach the model (for example, claiming a direct official connection while in fact routing through a third party) is a take-down offense.

3. Models you list

Each listing must name the model your endpoint actually serves.

• Models from other developers. List a model under its catalog name only if you serve that model unmodified. You must not list a fine-tuned, merged, distilled, or otherwise modified model under the original model's name, and you must not add hidden instructions intended to remove or weaken the model's safety behavior.
• Your own models. If you serve a model you trained, fine-tuned, merged, or otherwise modified, list it under your own model name and tell Onlist its base model and what you changed. Onlist shows the base model on the model page.

Onlist decides which models enter its catalog and may decline a model whose origin it cannot verify.

4. Prohibited services and public content

Onlist does not onboard or list services positioned as uncensored, unfiltered, jailbroken, or intended for sexual or adult content. You may not remove or weaken a model's safety behavior to enable content that the Terms of Service prohibit, and at a minimum every model you serve must refuse to produce sexual content involving minors.

Everything you publish on Onlist, including your store name, description, logo, and listing descriptions, must describe your service accurately and must not promote sexual or adult content, uncensored or unfiltered output, or ways to bypass safety measures. You may not promote your Onlist listings that way elsewhere either.

5. Data policy: separate "you" from "the upstream"

Data-policy declarations on Onlist are layered. Be specific about what you control:

• What you commit to as the listed provider. Whether your own infrastructure stores request or response bodies, for how long, and for what purpose (debugging, abuse investigation, none).
• What the upstream does. You may state your understanding (for example, that the model provider's commercial API contractually does not train on API traffic by default) but you may not present an upstream's policy as your own guarantee. If the upstream changes its policy, the accuracy of your declaration depends on you updating it.

You may not use prompts or outputs routed through your endpoint to train your own models, fine-tune adapters, or build derivative datasets, unless you obtain explicit informed consent from the requesting account holder out-of-band and disclose this practice in your declaration.

False or misleading data-policy claims may result in Onlist removing your endpoint from routing, withholding settlement, and terminating your account.

6. Reliability and abuse

You are responsible for endpoint availability, serving the traffic Onlist routes to you, protecting credentials, and preventing abuse on your side. Onlist may rate-limit, cool down, or temporarily stop routing traffic to your endpoint to protect buyers and platform integrity.

You must not silently substitute, downgrade, mislabel, or spoof models. If your own upstream changes capacity or behavior, update or pause your endpoint.

7. Pricing and settlement

You propose prices for your capacity subject to platform rules. Onlist meters successful usage and settles with you off-platform, deducting applicable platform fees, based on that usage. Settlement may be delayed, reversed, withheld, or denied for fraud, disputes, chargebacks, sanctions exposure, incomplete provider information, or violations of this Agreement. This settlement is a business arrangement between you and Onlist; the buyer's billing relationship is with Onlist.

8. Reviews and disputes

Verified buyers may review settled calls. You may dispute reviews or audit findings through Onlist tools, but you may not manipulate reviews, buy fake traffic, retaliate against reviewers, or otherwise interfere with platform signals.

9. Termination

You may pause or withdraw your endpoint at any time. Onlist may suspend or terminate your provider access for fraud, legal risk, safety issues, repeated outages, false declarations, or violations of this Agreement or the Terms of Service.

10. Contact

For the fastest response, sign in and submit a support ticket. You can also reach us at [email protected].