Scope and processing roles
This policy covers the Token Pyxis website, enterprise console and model API. Anyone may browse the website. Console and model API access is limited to contracted enterprises and their authorized users. Individual accounts and public registration are unavailable. Consult the enterprise contract for the provider’s legal name and contact details.
Enterprises generally determine the purposes and means of processing business data about their employees or end users. The gateway processes inputs required for configured model calls and handles relevant information for its own authentication, security and billing purposes. Roles and scope depend on the actual service and data processing agreement.
Personal information should be limited to what is necessary for specified and reasonable purposes, using the applicable legal basis. The responsible handler provides notice and obtains consent or separate consent where required.
Personal Information Protection Law, Articles 6, 13, 17 and 21
Data categories and purposes
- Accounts and sessions
- We process login identifiers, organizational membership, roles, password hashes and refresh credential hashes during account provisioning, member authorization and sign-in to authenticate users, manage sessions and enforce permissions.
- Model request metadata
- Request IDs, timestamps, key identifiers, models, endpoints, status, token usage and charges support metering, reconciliation, diagnostics and risk control.
- Inputs and outputs
- Prompts, attachments and other call inputs, along with model responses, are used to process and deliver calls and, if enabled on the instance, diagnose payload issues.
- Security and audit
- Depending on the feature, source IP addresses, failed authentication events, access restrictions and administration activity support risk detection, abuse prevention and review.
- Transactions and invoices
- Orders, wallet records, subscriptions and settlement data, plus invoice names, tax identifiers and contact details supplied when requesting invoices, support accounting and invoicing.
Upstream transfer, sharing and location
The gateway sends necessary inputs to the upstream provider selected by the model, enterprise settings and routing rules. Check provider terms and the enterprise agreement for processing scope, location and retention. When an optional feature such as web search is enabled, its queries may be sent to the enabled service.
The platform handles business payloads to carry out model calls, deliver responses and perform configured diagnostics. Entrusted processing, disclosure to independent handlers and disclosures required by law each follow applicable limits and obligations. Verify storage locations for the deployment and selected providers. Transfers of personal information outside China require the applicable notices, consents and procedures.
Personal Information Protection Law, Articles 21 and 23 and Chapter III
Retention and payload capture
Request metadata and diagnostic payloads are handled separately. The software defaults to 180 days of request-log retention, which deployment settings can extend. Where the six-month network log requirement applies, the deployment must configure and verify cleanup against that legal period. Account, transaction and financial records follow their respective purposes, agreements and retention duties.
Diagnostic payload capture is off by default. When enabled, it can record request payloads and optionally responses. Recorded payloads use AES-GCM at rest by default and follow separate retention settings with a 30-day software default. Confirm actual capture, scope, encryption and retention settings with your administrator. Turning capture off stops new records. Existing records are removed at expiry or through authorized cleanup.
PRC Cybersecurity Law, Article 23(3) · Data security controls
Access controls and browser storage
Role checks and tenant scoping restrict access, and key administrative actions are audited. Model API keys and console accounts use separate authentication. Authorized enterprise users can revoke or rotate keys and inspect related usage.
The console uses browser local storage for session continuity and cached account details. The website may also remember theme preferences. Sign out on shared devices and clear this site’s local data if needed.
Individual rights and requests
Individuals may exercise applicable rights to access, copy, rectify, supplement, delete, restrict or object to processing, withdraw consent, request an explanation of processing rules or request account closure. Mandatory records remain subject to applicable retention requirements after account closure or membership removal.
For requests involving an enterprise account, contact your enterprise administrator or the support channel in the agreement and include the account and requested action. Individuals can exercise applicable data rights without opening an enterprise account. We limit identity and authorization checks to what is necessary and handle verified requests under applicable law.
Contact and feedback · Personal Information Protection Law, Articles 44–50 · Network Data Security Management Regulations, Article 23
Authorized use and policy changes
The service is intended for enterprises and their authorized personnel. Processing personal information of children under 14 is subject to additional requirements, including guardian consent. Material changes to processing purposes, methods or scope require updated notice and renewed consent where required.
