About TokOne

TokOne is an enterprise-grade unified API gateway for AI models. We consolidate the compatibility protocols of OpenAI, Anthropic, Google, and other leading model providers behind a single platform, billed by actual token usage — sparing users the friction of juggling multiple provider accounts, cross-border payments, and network proxies.

Problems we solve

Enterprise teams hit four recurring blockers when adopting frontier models:

• Account barriers — overseas phone numbers, credit cards, and IPs; missing any one of them can get an account flagged or banned.
• Network instability — direct calls to provider endpoints suffer high latency and packet loss; self-built proxies bring compliance, security, and ops overhead.
• Vendor sprawl — invoices fragmented across providers, inconsistent billing units, painful reconciliation.
• Payments and invoicing — USD cards, long settlement cycles, no compliant invoices for corporate procurement.

TokOne handles all of this in a single gateway: full upstream protocol compatibility, high-speed direct connection, RMB settlement, real-time per-token billing, and compliant invoicing. If you're already using the OpenAI / Anthropic / Gemini SDKs, switching is a one-line base_url change.

Our commitments

100% genuine upstream models — guaranteed

Every request goes straight to the official upstream. We never silently swap models, downgrade quality, or rewrite system prompts. You can benchmark responses, pricing, and capabilities against the official APIs at any time — if there's a discrepancy, we'll compensate and disclose the cause.

Enterprise-grade data security

• Request bodies and model responses are not persisted. We only retain billing-required metadata: timestamp, model, token count, status code, and x-request-id.
• End-to-end TLS. API keys are read-only outside the console and cannot be exported.
• Per-group quota limits, per-IP allowlists, and per-rate / time-window controls.
• Log-minimization policy ensures: even in the event of a breach, conversation content cannot be reconstructed.

Fast, stable direct connection

Primary nodes are deployed across Japan, South Korea, Hong Kong, and other regions, with typical China-to-edge latency between 80–100 ms. BGP multi-line peering runs alongside the cross-border link — no VPN required for either enterprise networks or home networks. 99.9% monthly availability SLA, real-time incident announcements, and automatic traffic failover on link issues.

Transparent pricing, fixed exchange rate

Real-time per-token billing. Top up ¥7.14 to get $1 in platform credit at a fixed rate — RMB in, USD-denominated pricing for reconciliation, no exchange-rate drift. No monthly fees, no plan lock-in, no minimum spend.

Refundable balance

Balance never expires; unused credit is refundable through the original payment channel.

Pricing tracks upstream

When upstream providers adjust prices, we follow suit. Every change is announced in the changelog; calls already made are settled at the price in effect at the time.

Who it's for

• Enterprise R&D teams — needing a stable, auditable, invoice-ready multi-model backend, with unified procurement, usage monitoring, and cost attribution.
• AI product & agent builders — building long-running products that can't afford regional API restrictions or upstream jitter.
• AI coding tool users — running Cursor, Trae, Codex, Claude Code, Cline, CodeBuddy, and similar clients, wanting a one-time base URL switch onto a stable channel.
• Data-sensitive deployments — finance, legal, healthcare, government and enterprise environments that require non-persisted request bodies and end-to-end auditable invocations.

For integration, customization, or partnership inquiries, reach us via the console "Contact" button or our customer support channel.