Terms of Service

• API key management, model routing, usage statistics, cost details, team collaboration, and support.
• Supported models, pricing, quotas, and call records are shown in the console and model pages.
• Model capability, pricing, context length, rate limits, safety policy, availability, supported regions, and data handling rules may depend on the model provider and its applicable terms.
• Public pages do not promise fixed data regions, fixed model sources, fixed service levels, or permanent availability of any third-party model. Enterprise-specific capabilities require a separate agreement, console configuration, and product notice.
• This service is intended for development teams and enterprises outside mainland China and is not available in mainland China. CodexZH may restrict or deny access based on the region of the request origin (such as the IP location).

• You must ensure submitted data, prompts, code, files, and output usage are lawful, authorized, and non-infringing.
• You must protect accounts, passwords, API keys, access tokens, team permissions, and automation credentials.
• You are responsible for API calls, charges, model selection, permission changes, and other activity under your account, API keys, team members, or automated tasks.
• Team owners or enterprise administrators who provision, assign, review, or manage member accounts, subscriptions, usage, or API keys must provide necessary notices and obtain applicable authorization from those users.
• You remain responsible for business decisions, code releases, content publication, and downstream processing based on model outputs.

• Do not submit state secrets, work secrets, core data, important data without required procedures, unauthorized personal information, sensitive personal information, or minors' personal information.
• Do not evade laws, regulations, supervisory requirements, third-party provider authorization, access controls, content safety policies, rate limits, or cross-border data requirements.
• Do not use stolen, shared, traded, rented, or unlawfully obtained accounts, API keys, tokens, or credentials.
• Do not generate or distribute illegal, harmful, false, infringing, fraudulent, violent, sexual, discriminatory, malware, cyberattack, data theft, or phishing content.
• Do not reverse engineer, extract system prompts, model weights, training data, response patterns, or safety controls, or use jailbreaks, prompt injection, automated abuse, or rate-limit evasion to bypass protections.
• Do not use the service, outputs, response patterns, or model behavior to train, fine-tune, distill, evaluate, improve, or build competing AI models or substitute model services.

• Model outputs may be inaccurate, incomplete, biased, outdated, or unsuitable for a specific use case. You must evaluate outputs and remain responsible for downstream use.
• Legal, medical, financial, insurance, employment, housing, education testing, journalism, public affairs, and other rights-impacting or public-interest scenarios require human review, access control, and necessary disclosure.
• If you provide a chatbot, agent, or automated decisioning experience to external users, disclose that it is an AI system or AI-assisted system in an appropriate place.

Features labeled as preview, trial, beta, or experimental in pages, console, or communications may change, pause, or be removed and should not be used directly in production-critical workflows without assessment.

If CodexZH detects violations, abnormal calls, credential leakage, attacks, abuse, high-risk data submission, third-party provider rule violations, or data security risks, it may warn, rate-limit, filter, block, suspend, terminate, retain records, or cooperate with competent authorities for accounts, API keys, channels, requests, inputs, or outputs.

Provider price changes, rate limits, deprecations, policy changes, regional availability changes, exchange rates, or regulatory requirements may affect pricing, availability, or service policy. CodexZH will provide reasonable notice for material changes where possible, but urgent safety, compliance, upstream outage, or third-party service issues may require immediate action.