Terms
Terms of Service
Last updated: October 13, 2026
These terms govern access to and use of Zro.
Service
Zro provides API access to open-source model inference, account management, metered billing, and related operational tooling.
Acceptable use
You are responsible for the prompts, outputs, applications, and users connected to your account. Do not use the service for unlawful activity, abuse, credential sharing, attempts to disrupt infrastructure, or violations of applicable model, export, privacy, or safety requirements.
Accounts and plan use
Each account is for one person or one organization. You may not create or operate multiple accounts to obtain additional plan allowances, free trials, credits, or promotions. Plan allowances are for your own use. You may not resell or sublicense Zro inference access or API keys, share them with third parties, or offer Zro inference to others through another service, unless we have agreed in writing. Building your own application or agent on Zro is allowed; reselling raw Zro access is not. You may not sell, transfer, or assign an account. You may not create a new account to replace one that we have suspended or terminated. We may use payment, device, network, and usage signals to detect multiple, shared, replacement, or resold accounts. If we find a violation, we may suspend or terminate the accounts involved, revoke API keys, decline or reverse purchases, and withhold promotional credit. Except where applicable law requires otherwise, we are not obliged to refund fees or credits for an account that we suspend or terminate under this section.
Data handling
We process request content solely to provide inference. We do not retain prompts, responses, or request bodies, and the service provides no setting that enables retention. We do not use customer requests or logs for training, fine-tuning, evaluations, analytics, or other secondary purposes unless you separately and explicitly agree, and we do not authorize our providers to do so either. If we later offer a feature that stores request content, it will store content only if you turn it on, on the terms described in the Privacy Notice.
Inference capacity and providers
Zro fulfils inference requests using GPU capacity we operate and, where our own capacity is unavailable or saturated, capacity from third-party infrastructure providers engaged by Zro. We remain your counterparty and remain responsible for the service regardless of which capacity serves a given request. We select providers that maintain independent security certifications and offer GDPR-aligned data-processing terms, and we may add, change, or remove providers as capacity and compliance requirements evolve.
Regional processing
Every inference request passes through Zro's EU gateway for authentication and routing. Each API key specifies whether model inference may run in Europe, the United States, or either region. By selecting United States and creating or using that key, you instruct Zro to transfer inference request content to US infrastructure for processing. The region setting binds all inference capacity equally, including third-party providers: Zro will not intentionally route inference to a region that the key does not allow, and if no allowed deployment for the requested model is available from any source, the request may fail instead.
Region scope
The API-key region setting applies only to model inference, whether served by Zro or by an infrastructure provider. Account management, authentication, billing, operational analytics, optional web search, networking, and other service-provider processing may occur separately as described in the Privacy Notice. You are responsible for determining whether your selections and use of Zro satisfy your legal, regulatory, contractual, residency, and compliance obligations.
Billing
Usage-based billing is calculated from token usage and related billing metadata. You agree to pay fees associated with usage under your account and to keep payment information current.
Referral program
Eligible customers may invite new customers using their referral link. When a referral meets the requirements shown in the referral offer, Zro may provide promotional usage credit to the referring and referred customers. Referral credit may be used only for Zro usage, has no cash value, cannot be transferred or exchanged, and expires 90 days after it is issued. Self-referrals, duplicate or fictitious accounts, automated signups, spam, misleading promotion, and attempts to manipulate eligibility are prohibited. We may withhold or revoke credit, disqualify referrals, or restrict accounts in cases of abuse or a disputed qualifying payment. We may change, suspend, or end the referral program for future referrals.
Availability
The service may change as models, infrastructure, dependencies, or compliance requirements evolve. We may suspend access to protect the platform, users, or third parties.
Prohibited uses
You agree not to use Zro, its APIs, CLI tools, documentation, or any models made available through the service (including standard, unaligned, or security-oriented models) to generate, transmit, facilitate, process, or assist in any of the following: Child Exploitation and Abuse: Generating, storing, processing, or disseminating Child Sexual Abuse Material (CSAM), Child Sexual Exploitation and Abuse (CSAE), or any content depicting or promoting the exploitation of minors. Weapons of Mass Destruction and Public Harm: The design, development, production, stockpiling, or proliferation of chemical, biological, radiological, or nuclear (CBRN) weapons, explosives, or materials intended to cause mass casualty or severe physical injury. Unauthorized Cyber Operations and Malicious Attacks: - Conducting unauthorized network intrusions, denial-of-service (DoS/DDoS) attacks, automated vulnerability exploitation, port scanning, credential stuffing, or phishing. - Generating malware, ransomware, weaponized exploits, or payload components, except when developed and utilized strictly for authorized defensive security research, vulnerability verification (PoC), or controlled validation within systems you own or are explicitly authorized to test. - Scanning, attacking, evaluating, or pen-testing any network, system, application, or software without explicit, verifiable, and documented authorization from the verified system owner. Terrorism and Violent Extremism: Facilitating, coordinating, or inciting acts of terrorism, violent extremism, or targeted physical harm against individuals, organizations, or critical infrastructure. Platform and Infrastructure Abuse: Attempting to bypass service rate limits, reverse-engineer proprietary deployment kernels, access unauthorized tenant environments, or interfere with the integrity, availability, or operational performance of the Zro network and multi-region infrastructure. Downstream and End-User Compliance: If you integrate Zro into third-party agents, applications, or downstream services, you are solely responsible for ensuring that your end users do not engage in any activities prohibited under this Section. Enforcement and Remedies: While Zro does not inspect, retain, or monitor prompt or completion contents by default, Zro reserves the right to immediately suspend or terminate access, revoke API keys, and forfeit unused credits without prior notice upon receipt of credible third-party abuse reports, formal legal notices, law enforcement requests, or upon detecting unauthorized activity, infrastructure abuse, or fraudulent conduct associated with an account. In cases involving severe illegal conduct, Zro may report associated account metadata to law enforcement as required by law.
Uncensored security models
Some models offered through Zro, including dolly1-security, are unaligned or security-oriented and do not apply the standard safety filters, refusals, or moderation layers found in general-purpose models. They may generate inaccurate, offensive, or high-risk content, and their outputs do not constitute professional security, legal, or compliance advice. You are solely responsible for your use of these models and their outputs. You represent and warrant that your use is lawful, adheres to applicable cybersecurity regulations, and is performed exclusively on systems you own or have explicit, documented authorization to evaluate. To the maximum extent permitted by applicable law, you assume all risks and agree to defend, indemnify, and hold harmless Zro, its affiliates, and its operators from any claims, damages, liabilities, or losses arising from or related to your use of these models or their outputs.
Disclaimer of warranties and limitation of liability
The service, APIs, and model outputs are provided on an "as is" and "as available" basis without warranties of any kind, whether express or implied, including warranties of merchantability, fitness for a particular purpose, or non-infringement. We do not warrant that outputs are accurate, error-free, or compliant with specific legal requirements. To the maximum extent permitted by applicable law, in no event will Zro, its operators, or affiliates be liable for any indirect, incidental, special, consequential, or punitive damages, or any loss of profits, data, or business interruption. Our total aggregate liability arising out of or related to your access to or use of the service will not exceed the total fees paid by you to Zro under these terms in the twelve (12) months preceding the incident giving rise to liability.