The below terms (the “Preview Terms”) apply to your participation in the Public Preview of, and your use of, the Model Evaluation 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.
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.
The Preview Offering provides access to tools for testing, evaluation, and feedback purposes. This offering is intended to assist users in assessing model performance through the Model Evaluation Service.
2.1 Preview Pricing: As specified in the Gradient AI Platform Pricing page during the Preview Period. DigitalOcean reserves the right to change pricing upon 30 days’ notice.
2.2 Support: No formal Technical Support or Service Level Agreements (SLAs) apply. Support is provided on a reasonable-effort basis during standard business hours.
3.1 Your use of the Preview Offering is subject to all applicable third-party requirements and the DigitalOcean Acceptable Use Policy. Users are strictly prohibited from utilizing the Service for any purposes that violate applicable laws or regulations, including the generation of harmful content or malware.
3.2 The Model Evaluation Service utilizes the DeepEval open-source framework and models from third-party AI providers (including the Judge Model). Your use of these components is subject to the AI Model Terms, in addition to these Preview Terms. You are responsible for ensuring your use of the Service complies with all applicable third-party license terms.
As a Beta Service, the Preview Offering may not meet the same reliability, performance, or security standards as generally available Services. You are solely responsible for determining whether and how to use the Preview Offering and should validate it in your environment before relying on it for production workloads.
5.1 Customer Evaluation Datasets: Data submitted for evaluation and the resulting Evaluation Outputs (results and metrics generated by the Service) are deemed your Services Content, owned exclusively by you. For the purpose of these Preview Terms, this Services Content is deemed your Confidential Information.
5.2 Datasets. You represent that you possess all necessary rights, licenses, and permissions to provide the Evaluation Datasets to DigitalOcean for the purposes described in these Preview Terms, and that our use of such datasets will not violate any third-party rights or applicable laws.
5.3 Model Training: We will not use Evaluation Datasets or Evaluation Outputs to train, fine-tune, or improve any of our underlying models, including the Judge Model or any other DigitalOcean-provided.
5.4 Judge Model Performance: The Model Evaluation Service relies on a ‘Judge Model’ and probabilistic methods to provide performance estimations. Evaluation Outputs are not a guarantee of production performance, suitability, or legal compliance.
5.5 Human Oversight: You acknowledge that the Judge Model is not perfect, and the Evaluation Outputs are intended to inform, not replace, mandatory human oversight and review before making production deployment decisions.
5.6 Safety and Bias Metrics: Metrics related to safety and bias are directional tools. They do not constitute a certification or guarantee that a model is legally “safe,” “compliant,” or meets specific regulatory requirements in any jurisdiction.
5.7 Storage and Fees: Evaluation Datasets and Outputs are stored within the platform, currently at no cost, to facilitate comparative analysis. We reserve the right to introduce storage charges or enforce a retention policy (e.g., a time-to-live) in the future, upon reasonable notice.
5.8 Data Responsibility: You are solely responsible for ensuring that Evaluation Datasets are de-identified and do not contain sensitive data or Personally Identifiable Information (PII) before upload.
5.9 Security and Privacy: Your Evaluation Datasets and Evaluation Outputs are subject to our security and privacy requirements, including protections against cross-tenant leakage and unauthorized access.
If you use your own model (a ‘Candidate Model’) for evaluation, you represent and warrant that you possess all necessary commercial rights, licenses, and authorizations (including for attribution, if required) to use and evaluate the model and permit DigitalOcean to host and run the Candidate Model for the purpose of the Preview Offering.
7.1 Accuracy: Outputs generated by AI models may be inaccurate, incomplete, or biased. You must independently verify all outputs before relying on them.
7.2 No Fiduciary Duty: The Preview Offering is a tool; it does not provide professional advice (legal, medical, or otherwise).
DigitalOcean may suspend or terminate your access to the Preview Offering at any time (in our sole discretion), with or without notice, if we choose to discontinue the service or if we believe your use violates these Preview Terms. Upon the effective date of termination or suspension of the Preview Offering, DigitalOcean will delete all Evaluation Datasets and Evaluation Outputs in accordance with our standard data retention and deletion policies.
AS A BETA SERVICE, THE PREVIEW OFFERING IS PROVIDED “AS IS” AND “AS AVAILABLE”. TO THE MAXIMUM EXTENT PERMITTED BY LAW, DIGITALOCEAN DISCLAIMS ALL WARRANTIES, AND OUR TOTAL AGGREGATE LIABILITY FOR ANY CLAIMS ARISING FROM THE PREVIEW OFFERING WILL NOT EXCEED $100.