Last updated: October 1, 2026
1. Who we are and what these terms cover
These Terms of Service ("Terms") govern your use of inferbase.ai, the Inferbase API at api.inferbase.ai, the dashboard, and the documentation (together, the "Service"). The Service is operated by Inferbase, based in Bengaluru, India ("Inferbase", "we", "us"). Inferbase is not yet incorporated; when it is, this page will name the company that operates the Service.
By creating an account, creating an API key, or otherwise using the Service, you agree to these Terms, our Acceptable Use Policy and our Privacy Policy. If you use the Service on behalf of an organization, you confirm you are authorized to bind it, and "you" means that organization.
If you have a signed agreement with us (for example an Enterprise order form), that agreement prevails over these Terms where they conflict.
2. The Service
Inferbase is a gateway that sits between your application and AI models. Through one OpenAI-compatible API it:
- routes each request to a model, either the one you name or one chosen for the task under your policy ("routing");
- applies the policies you configure: model pools, budgets, rate and spend limits, guardrail checks, and tool rules ("policies");
- serves the request, either on our own accounts with model providers ("managed inference") or on credentials you connect: your provider API keys ("your keys") or your cloud accounts with Microsoft Azure, Amazon Web Services or Google Cloud ("your cloud accounts");
- records each request's routing decision, usage and cost ("logs").
We also publish a model catalog, comparison and recommendation tools, and GPU capacity tools ("catalog tools"), which are free to use and do not require an account.
We may change, add or remove features. If we remove a feature you pay for, we will give you reasonable notice where we can.
3. Accounts, organizations and API keys
3.1 You must give accurate information when you register and keep it current. You must be at least 16 years old, or the age of digital consent where you live if that is higher.
3.2 Each account belongs to an organization. An organization's owners and admins control its members, projects, keys, policies, connected credentials and billing, and can see its usage and audit log. You are responsible for everything done under your organization, including by its members and through its API keys.
3.3 Keep your password, authenticator and API keys confidential. Tell us at once at legal@inferbase.ai if you believe one has been compromised. You can revoke and rotate keys in the dashboard at any time.
3.4 You may close your account at any time from account settings. Closing it deletes it as described in the Privacy Policy. Unused prepaid credit is handled under section 8.6.
4. Your content
4.1 "Your content" means what you send through the Service (prompts, messages, files, images, audio, tool definitions and tool results) and what the Service returns to you (outputs). As between you and us, you own your content.
4.2 You give us a limited licence to host, copy, transmit and process your content only to provide the Service to you: to route and serve requests, to apply your policies, to keep the logs and records described in the Privacy Policy, to operate and secure the Service, and to measure and improve routing quality as described there. We do not sell your content and we do not use it to train AI models.
4.3 Request content is stored for 30 days by default so you can inspect requests and we can verify routing. You can turn this off for yourself, for your organization, or for any API key, and you can delete stored content at any time. The Privacy Policy describes this in full.
4.4 You are responsible for your content and for having the rights and consents needed to send it to us and to the models you use, including for any personal data in it. Do not send us special categories of personal data, payment card data or health records unless you have a signed agreement with us that covers them.
4.5 Outputs are generated by AI models we do not build. They can be wrong, incomplete, biased or unsuitable, and similar outputs may be given to other users. You must evaluate outputs before relying on them, and you must not represent an output as human-generated where that would mislead.
5. Models, providers and your credentials
5.1 Managed inference. When a request is served on our accounts, we send it to the model provider that serves the chosen model. The providers we use are listed on our subprocessors page and may change. Model creators and providers may impose their own use policies and licence terms for a model (for example a model's licence or a provider's acceptable use policy); you agree to comply with the terms of the models you use, which our Acceptable Use Policy incorporates.
5.2 Your keys and your cloud accounts. When you connect a provider API key or a cloud account, we act on your instructions to send your requests to that provider or cloud under your account. Your agreement with that provider or cloud governs its processing, its pricing and its billing, and you remain responsible for complying with it. We store keys and cloud credentials encrypted and use them only to serve your requests, to list and check the models they reach, and to show you their status. Where you connect a cloud account through an IAM role or workload identity, you control that trust and can revoke it in your cloud at any time.
5.3 MCP servers, webhooks and other endpoints you configure. If you register an MCP server, a webhook endpoint or another endpoint, we will connect to it and send it data on your instruction. You are responsible for those endpoints and for what they do with the data.
5.4 Models are added, changed, repriced and retired by their creators and providers, sometimes without notice to us. We may remove a model from routing or from the catalog when it stops being servable. We do not guarantee that any particular model will remain available.
5.5 Routing chooses among models eligible under your policy using the information we have at the time, and records why. Routing is a best-effort decision, not a guarantee that the chosen model is the best for your request.
6. Acceptable use
You must follow our Acceptable Use Policy, which forms part of these Terms. In summary, you must not use the Service to break the law, to harm people, to abuse or attack the Service or others, to circumvent our limits or safety measures, or to break the terms of the models and providers you use. We may block requests, keys or accounts that do.
7. Plans and limits
7.1 We offer the plans described at inferbase.ai/pricing, currently Free, Pro and Enterprise. Each plan includes a number of requests each calendar month (UTC) and may include daily limits on managed inference, rate limits and feature limits. A request counts toward your plan when it succeeds, whether it is routed or pinned and whether it runs on managed inference, your keys or your cloud accounts.
7.2 On the Free plan, managed inference is free within the published limits. We may change those limits, and we may suspend free managed inference where it is abused.
7.3 When you reach a plan limit, further requests are refused until the limit resets, you upgrade, or (on Pro) you buy additional requests as described on the pricing page, up to any cap you set.
8. Fees and payment
8.1 Paid plans and credit are not yet available for self-serve purchase. This section applies once they are, and to any Enterprise agreement that does not provide otherwise.
8.2 Plan fees. Pro is billed monthly in advance. Additional requests beyond the included amount are billed in blocks at the published price, up to the cap you set.
8.3 Managed tokens. On paid plans, managed inference is charged per unit the provider bills (tokens, images, seconds of audio or video) at the provider's price plus the published markup, from prepaid credit you add. Each request's receipt shows its charges. Guardrail checks that call a model are charged the same way, including on your own keys.
8.4 Your keys and cloud accounts. We do not charge for tokens served on your own keys or cloud accounts; your provider bills you for them. Requests still count toward your plan.
8.5 Payment processing. Payments are processed by the payment provider named at checkout, which may act as the merchant of record, in which case its terms apply to the transaction, including taxes, invoices and receipts.
8.6 Refunds. Plan fees and prepaid credit are non-refundable except where the law requires a refund or where we terminate the Service without cause, in which case we refund unused prepaid credit.
8.7 Changes. We may change prices with at least 30 days' notice by email or in the dashboard. A change takes effect at your next billing period after the notice period.
8.8 If your credit reaches zero, managed requests that need credit are refused until you add more. Holds placed on credit while a request runs are released or settled when it finishes.
9. Our intellectual property
The Service, its software, routing, documentation, design and our original catalog content belong to us or our licensors. We give you a limited, non-exclusive, non-transferable right to use the Service under these Terms. Catalog data about third-party models (specifications, prices, benchmarks) comes from their creators, providers and public sources; we do not claim ownership of it and do not guarantee it is accurate or current. Feedback you give us may be used without obligation to you.
10. Suspension and termination
10.1 We may suspend or restrict an account, organization or key, immediately and without notice where reasonably necessary, if: it breaches these Terms or the Acceptable Use Policy; it creates a security, legal or financial risk to us, our providers or others; payment fails; or a provider or law requires it. Where practical we will tell you why and how to fix it.
10.2 We may end these Terms for convenience with 30 days' notice. You may stop using the Service at any time.
10.3 On termination your right to use the Service ends, and we delete your data as described in the Privacy Policy. Sections 4.5, 9, 11, 12, 13 and 15 survive.
11. Disclaimers
THE SERVICE, THE CATALOG TOOLS AND ALL OUTPUTS ARE PROVIDED "AS IS" AND "AS AVAILABLE". TO THE EXTENT THE LAW ALLOWS, WE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY AND NON-INFRINGEMENT. WE DO NOT WARRANT THAT THE SERVICE WILL BE UNINTERRUPTED OR ERROR-FREE, THAT ANY MODEL OR PROVIDER WILL BE AVAILABLE, OR THAT GUARDRAIL CHECKS WILL CATCH ALL HARMFUL CONTENT. No uptime commitment applies unless an Enterprise agreement states one.
12. Limitation of liability
12.1 TO THE EXTENT THE LAW ALLOWS, NEITHER PARTY IS LIABLE FOR INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE DAMAGES, OR FOR LOST PROFITS, REVENUE, DATA OR GOODWILL.
12.2 TO THE EXTENT THE LAW ALLOWS, OUR TOTAL LIABILITY ARISING OUT OF OR RELATING TO THE SERVICE IS LIMITED TO THE GREATER OF THE AMOUNTS YOU PAID US IN THE 12 MONTHS BEFORE THE EVENT GIVING RISE TO THE CLAIM AND USD 100.
12.3 Nothing in these Terms limits liability that cannot be limited by law, including for fraud, or for death or personal injury caused by negligence.
13. Indemnity
You will defend and indemnify us against third-party claims, and the resulting losses and reasonable costs, arising from your content, your use of outputs, your breach of these Terms or the Acceptable Use Policy, or your breach of an agreement with a provider or cloud you connect.
14. Changes to these Terms
We may update these Terms. We will post the new version here with its date and, for material changes, notify account owners by email or in the dashboard at least 15 days before they take effect, except where a change is required by law or to address abuse. If you do not agree, stop using the Service before the change takes effect.
15. Law and disputes
These Terms are governed by the laws of India. The parties will first try to resolve a dispute by good-faith negotiation for 30 days. A dispute that remains will be resolved by arbitration under the Arbitration and Conciliation Act, 1996, by a sole arbitrator, seated in Bengaluru, in English. Either party may seek urgent interim relief from the courts of Bengaluru, Karnataka. If you are a consumer, this does not take away rights the law where you live gives you.
16. General
These Terms, the Acceptable Use Policy, the Privacy Policy and any order form or DPA you sign are the whole agreement between us about the Service. If a provision is unenforceable, the rest remains in effect. A failure to enforce a provision is not a waiver. You may not assign these Terms without our consent; we may assign them to an affiliate or in a merger or sale of the business. Neither party is liable for delays caused by events beyond its reasonable control, including outages of model providers or clouds.
17. Contact
Legal and security: legal@inferbase.ai. Privacy: privacy@inferbase.ai. Inferbase, Bengaluru, India.