Let every tokenflow into the city
t.city — a compliant large-model platform and unified service gateway for mainland China. Pooling domestic compute into one access, scheduling and billing layer, so AI runs as readily as water and power.
- 300+
- Enterprises
- 20+
- Models
- 99.9%
- Uptime
- <50ms
- Routing
One entrance to the whole city of intelligence
Stop wiring a separate API per model. t.city orchestrates compute and models the way a city organizes itself — integrate once, then route across mainland open and open-weight models on demand, with unified billing and centralized governance.
- One unified API
- Multi-model smart routing
- Unified quota & billing
- Central observability & audit
Compliance written into the foundation, not just promised
t.city's mainland site is operated and ICP-filed by 唐马供销有限公司 (Tangma Supply & Marketing Co., Ltd.) — the filing entity is the actual operator — together with mainland-China AI compute centers, on a principle of region-based compliant delivery: mainland workloads run on domestic open and open-weight models, with private deployment, local inference, and data that never leaves the domain.
Data stays in-domain
Local inference; sensitive data never leaves the mainland or transits third parties.
Graded-protection ready
Designed and delivered for China's classified-protection and sector regulation.
Domestic models first
Prioritized access to GLM, DeepSeek, Qwen and other domestic models.
Auditable & traceable
Every call, quota and charge is logged for audit and compliance review.
Compliance terms are governed by the signed service contract; this page is descriptive and not legal advice.
From access to governance — a city's full infrastructure
Unified model access
One interface across mainland open & open-weight models.
Smart routing
Auto-select the best model and compute by cost, latency and policy.
Private deployment
One-click private deploy; full control of models and data.
Enterprise console
Quotas, keys, members and usage, all in one place.
Observability & audit
Live usage, success rates, end-to-end logs and alerts.
Elastic compute
Burst across multiple compute centers on demand.
One city, three districts
t.city weaves the unified gateway, the compute backbone and compliant delivery into one city of three districts — a single token flowing seamlessly across all three.
An AI compute network rooted in mainland China
t.city unites AI compute centers across regions into one schedulable grid — nearby access, elastic scaling, and stable supply.
The city is already running
Model families integrated
Three steps to wire the whole city into your product
Apply
Submit your needs and compliance scope; get a tailored plan.
Integrate
One API — configure model routing and quotas.
Go live
Monitoring, billing and audit — continuously operated.
Ready to move your tokens into the city?
Join t.city — a compliant, unified, scalable model gateway.