Compliance & Acceptable Use

1. What ModelAPI Is (and Is Not)

• ModelAPI is an AI algorithm and orchestration cloud: Harness Control, Nexus routing, session budgets, context compaction, semantic cache, and observability — plus licensed upstream inference.
• We bill at official model rates plus a transparent infrastructure fee. Algorithm and orchestration layers are our proprietary software services.
• We do NOT operate reverse-engineered web scraping, stolen API pools, subscription-to-API (Sub2API), or identity-spoofing proxies.
• We do NOT guarantee availability in jurisdictions where upstream providers prohibit use.

2. Who May Use ModelAPI

• Companies and organizations legally registered outside mainland China (PRC), using the service in compliance with local law and upstream provider Terms.
• Individual developers residing in jurisdictions permitted by each model provider's Terms of Service (e.g. Anthropic, OpenAI, Google, AWS acceptable use policies).
• NOT eligible: entities registered in mainland China (companies, institutions, organizations), or accounts operated primarily on behalf of such entities.
• NOT eligible: users whose primary purpose is to circumvent export controls, sanctions, or provider geo-restrictions in violation of applicable law.

中文界面提供微信/支付宝等支付方式，仅为符合条件的海外或允许辖区用户提供便利，不构成向受限主体提供服务的邀请。

3. Your Data — Your Responsibility

• You retain ownership of prompts, files, and outputs subject to upstream provider Terms.
• You are solely responsible for: legality of content; classification of personal data, trade secrets, and regulated data; cross-border transfer compliance (PIPL, GDPR, etc.); obtaining consents and legal bases.
• Do NOT send state secrets, unauthorized personal information, credentials, or content that violates law or third-party rights.
• ModelAPI processes requests to route inference and for billing, abuse prevention, and security. We do not use your API content to train our own models.
• For sensitive workloads, use BYOK (bring your own provider key) and minimize data sent through the gateway.

4. Acceptable Use

• Illegal content or activity: fraud, CSAM, terrorism, hate, harassment, malware, unauthorized access.
• Using the API to assist crimes (including telecom fraud, money laundering) — zero tolerance.
• Reselling or sublicensing API access without written authorization.
• Automated abuse: credential stuffing, DDoS, scraping provider infrastructure.
• Circumventing provider safety filters or moderation without authorization.

我们有权暂停或终止违规账号，并在法律要求时配合监管与司法机关。详见 服务条款。

5. Prohibited industry practices (we do NOT offer these)

6. Cross-border data & export control

• If your prompts contain personal information or regulated data, you must comply with PIPL, DSL, GDPR, and other applicable cross-border rules — including security assessments where required.
• ModelAPI routes inference to upstream regions (e.g. AWS us-east-1). You choose what to send; minimize sensitive payloads.
• Anthropic, OpenAI, Google, and AWS publish geo and entity restrictions (including policies on China-controlled entities and relay intermediaries). You must comply; we may suspend accounts that violate upstream or sanctions rules.
• We do not provide legal clearance for your use case. Consult qualified counsel before processing regulated data.

7. Product positioning

• ModelAPI sells software: agent orchestration (Harness, Nexus), metering, dashboards, and support — not a simple token relay or circumvention tool.
• Upstream inference uses licensed B2B channels (Bedrock, Vertex/Gemini, etc.). Our margin is infrastructure + agent orchestration services, not hidden token markup.
• Harness profiles (claude-code-optimal, hermes-terminal, openclaw-trader) are server-side orchestration products documented at /docs#harness.
• Enterprise customers may use BYOK, spend caps, and invoicing for clearer contractual alignment.
• Payment convenience (WeChat/Alipay) does not expand eligibility beyond this policy.

8. Upstream Provider Terms

使用特定模型即表示您同时受该模型厂商及云服务商条款约束，包括但不限于：

• Anthropic Acceptable Use Policy & Commercial Terms (via AWS Bedrock or direct BYOK)
• OpenAI Usage Policies & API Terms
• Google Cloud / Gemini API Terms
• AWS Acceptable Use Policy & Bedrock terms
• DeepSeek and other listed providers' policies

9. Industry context

10. Contact

合规与滥用举报： compliance@aimodelapi.ai
一般支持： support@aimodelapi.ai