The below terms (the “Preview Terms”) apply to your participation in the Public Preview of, and your use of, the above identified DigitalOcean Service (the “Preview Offering”). The Preview Terms supplement the terms of the DigitalOcean Terms of Service Agreement, including Service Terms, or such other written agreement between you and DigitalOcean governing your use of our Services (our “Agreement”). The Preview Offering is a Beta Service for purposes of the Agreement, and terms not defined in these Preview Terms have such meaning otherwise defined in the Agreement. If there is a conflict between these Preview Terms and the Agreement, these Preview Terms will control solely with respect to the conflict relating to your use of the Preview Offering.
Subprocessor Disclosure: DigitalOcean uses Amazon Web Services, Cloudflare, and Traversal (located in the United States) as subprocessors to support delivery of the Preview Offering service. A current list of other DigitalOcean Service subprocessors is available here: https://www.digitalocean.com/trust/subprocessors.
Your participation in the Public Preview and use of the Preview Offering is voluntary. If you do not accept these Preview Terms then you may not participate in the Public Preview.
1. Scope & Service Status
a. Description. The Preview Offering is an API-first router for AI model inference. When you send a request to the router’s model name, the router assigns the request to a use case and selects, from a pool of Third-Party Models (some hosted by DigitalOcean), the model with the highest benchmark score for that use case, adjusted for any Routing Preference you set (Section 1.4) or personalization you request (Section 5). The Preview Offering is a limited-availability release designed for testing and feedback and is not a General Availability (GA) service.
b. Relaunch changes. Effective October 12, 2026, DigitalOcean is relaunching the Preview Offering as an intelligent router. As part of that change: (a) preset routers, the create-router flow, Router Tasks, custom tasks, and fallback configuration are removed and (b) requests previously routed by a preset or custom router may be served by different models, with different pricing, performance, providers, and outputs. Continued use of the Preview Offering after the effective date means you accept these Preview Terms and the changes described in this Section.
c. Third-Party and Open-Weight Models. Models in the pool are Third-Party Services or open-weight models. Your use of the Preview Offering is subject to the DigitalOcean Terms of Service Agreement, Service Terms (including the AI Model Terms), and to each model provider’s license, acceptable use policy, and other terms (together, “Model Terms”), which are linked from the Model Catalog. Because the router, not you, selects the model, you agree to the Model Terms for every model in the pool when you use the router and are responsible for complying with them, including any attribution requirements, for any model that serves your request. DigitalOcean does not develop, control, endorse, or warrant any pool model and is not responsible for its accuracy, availability, or security. The “Managed by DigitalOcean” label refers only to DigitalOcean’s operation of the Preview Offering and implies no affiliation with or endorsement by any model provider. DigitalOcean may share information about your use of a third-party model (excluding your Services Content) with that model’s provider as described in the Service Terms.
d. Benchmarks and Routing. Benchmark results, scores, and cost figures published in the Inference Router Overview or Model Catalog are illustrative only. They are produced under specific conditions, may be provider-reported or measured by DigitalOcean, change over time, and may not predict results for your data or workload; they are not a representation, warranty, or guarantee of accuracy, quality, speed, cost, or savings. Benchmark and model names belong to their owners, are used only for identification, and imply no endorsement. Published benchmark and pool information is not Beta Service Confidential Information. Any routing preference biases model selection but does not guarantee any particular cost, quality, or latency.
e. No Service Level Agreements (SLAs): There are no uptime, performance, or availability guarantees. Existing DigitalOcean SLAs do not apply.
2. Customer Acknowledgments
Customer Acknowledges it will:
a. Not run sensitive workloads. We expect but do not guarantee that Public Previews perform for production-level workloads. We recommend using simulated test data and not running sensitive workloads on Public Preview Services.
b. Provide feedback at the completion of the Public Preview period.
3. Pricing & Billing
a. No routing charge. The Inference Router service is provided at no additional charge during this phase of the Public Preview. DigitalOcean will provide advance notice before billing for the router itself begins.
b. Model charges. Each request is billed at the then-current rate of the model that served it, as shown in the Model Catalog. Rates differ between models, and no routing preference or benchmark figure guarantees lower charges than another preference or than calling a specific model.
c. Underlying resources. Use of underlying inference resources (e.g., GPU Droplets, Dedicated Inference) will continue to be billed at their standard rates.
4. Confidentiality and Beta Participation
Notwithstanding any broader definition in the Agreement, “Beta Service Confidential Information” includes your feedback and any non-public benchmarks, test results, or observations about the Preview Offering, but excludes (a) information published under Section 4 and (b) your Evaluation Data and Services Content, which are governed by Section 5 and the Terms of Service Agreement. You may use Beta Service Confidential Information only to evaluate the Preview Offering internally and may not disclose or publish it, including in comparisons or benchmarks, without DigitalOcean’s written consent.
5. Changes to the Preview Offering
DigitalOcean may add, change, or remove features, use cases, benchmarks, pool models, console functions, and APIs of the Preview Offering, and may suspend or end the Public Preview or your access, at any time. Where a change materially affects existing functionality, DigitalOcean will use reasonable efforts to give notice by email to your account contacts or in the console. Configurations may be deleted when the Public Preview ends; export anything you need beforehand.
6. Support & Feedback Channels
Support is provided on a reasonable-effort basis through your account team during standard business hours (8:00 AM – 5:00 PM PT, Monday through Friday, excluding holidays).
7. Ending the Public Preview
a. You may disable the Preview Offering at any time by reaching out to Customer Support.
b. DigitalOcean makes no promise that the Preview Offering will continue to General Availability.
c. Upon release of a future version (if any), these Preview Terms will automatically terminate, and continued use will be governed by the Agreement and any updated Service Terms.