Why Niu? A Token Management Platform for Developers and Teams
Performance, reliability and economy for people using model APIs. Access models, manage token spending and understand every request.
Developers and teams use model APIs to build applications, work with coding tools, and explore ideas in Chat and Video. As usage grows, they need more than another API key: they need to understand which models they can use, what happened to a request, and where their token budget went.
Niu is a token management platform for those users. Its focus is performance, reliability and economy: access models efficiently, understand failures, and keep spending visible and controlled.
Performance: keep model access efficient
Use documented OpenAI-compatible and Anthropic-compatible endpoints to connect applications and coding tools through a workspace API key. Check the model’s supported capabilities before switching an application.
The Rust gateway supports streaming and has a published gateway benchmark. Those measurements describe historical versions against a deterministic mock upstream. They do not promise the same end-to-end latency or throughput for every model and workload.
Reliability: understand each request
Request history brings status, recorded attempts, reported token usage, timing and charges together. You can distinguish a request rejected before dispatch from an upstream error or an interrupted response.
Configured model routes support bounded failover. An uncertain request must not be blindly replayed: doing so could repeat work or incur another charge. Missing usage evidence remains explicit. See the request evidence guide for supported behavior and limits.
Economy: make token spending visible
Review listed model prices, actual token usage and request charges. Set workspace and API-key budgets for applications and team usage. In-flight reservations count toward applicable spending limits before requests finish.
Economy means making informed choices and keeping spending under control. It is not a claim that every model is cheaper. Model prices and availability depend on configured offers, and unresolved billing evidence must remain distinguishable from a confirmed charge.
One workspace for API, Chat and Video
Call APIs from your applications, compare models in Chat, and revisit saved conversations. Supported video workflows expose saved job state and authorized result access. Availability depends on model configuration and permissions.
Niu’s open-source gateway, dashboard, catalog and Docs can also be self-hosted with PostgreSQL. See the deployment guide and release acceptance matrix for setup and qualification boundaries.
Browse models or follow the quickstart to get started.