Acceptable Use Policy

This Acceptable Use Policy ("AUP") sets out what you can and cannot do with the clusterbid Service. The AUP applies regardless of which model, region, or SKU you use. It applies in addition to the Terms of Service; AUP violations are grounds for suspension or termination.

You must not use the Service to generate, distribute, or facilitate:

• CSAM — child sexual abuse material in any form. Zero tolerance; reported to NCMEC and the equivalent regional authorities.
• Weapons of mass destruction — content meaningfully advancing the development or deployment of chemical, biological, radiological, nuclear, or high-yield explosive weapons.
• Cyberattacks — exploit code, malware, phishing kits, or credential-stuffing infrastructure targeting systems you do not own or have written authorization to test.
• Targeted harassment — content designed to harass, threaten, or dox specific individuals.
• Sexual content involving minors or non-consenting adults.
• Election interference — coordinated impersonation of officials, fake voting infrastructure, or content materially designed to suppress legitimate voting.
• Fraud and deception at scale — operating fraudulent businesses, generating fake reviews, or deceptive impersonation of real entities or individuals.
• Illegal activity in any jurisdiction in which you operate, including controlled-substance synthesis instructions, human trafficking, and illegal arms trade.

Some applications are not prohibited but require you to operate a meaningful safety layer on top of the API. We expect a documented review process and human oversight for:

• Medical, mental-health, and clinical decision support.
• Legal advice provided to end users.
• Financial advice or automated trading decisions.
• Employment, housing, credit, or insurance decisions that affect individuals (in-scope of the EU AI Act high-risk classification).
• Critical-infrastructure operations.

If you are running any of these, we recommend Dedicated deployments and a contract conversation before going to production.

Hermes-3 and any future agent-tuned model in the catalog are intentionally more steerable than the standard instruct variants. The AUP applies in full regardless of steerability. Customers running agent-tuned models are expected to operate their own moderation and guardrail layer; if you are not, you should use an instruct model.

The Service is subject to per-account rate limits. Soft throttle precedes hard 429s. Limits scale automatically with verified billing history.

Attempts to circumvent rate limits — including using multiple accounts to evade limits on a single workload, automated reCAPTCHA defeat against the signup flow, or stolen payment methods — are AUP violations and grounds for immediate termination.

• Do not probe, scan, or test the vulnerability of the Service without prior written authorization. Pen-testing is permitted on Dedicated under an LoA — email security@clusterbid.eu.
• Do not interfere with or disrupt other customers' use of the Service.
• Do not attempt to extract model weights, training data, or other proprietary information beyond what the API returns by design.
• Do not use API responses to train competing inference services without our written agreement.

We may, at our discretion: warn you, restrict capabilities, throttle, suspend, or terminate access. Suspension notices cite the AUP clause and an internal case ID. Customers may appeal to appeals@clusterbid.euwithin 30 days.

For zero-tolerance categories (CSAM, WMD content, infrastructure-level cyberattack tooling), enforcement is immediate termination with no appeal and may include law-enforcement referral.

To report suspected abuse or AUP violations, email abuse@clusterbid.eu. Include the API key prefix, timestamps, and a description of the issue. We respond within 1 business day and faster for active incidents.