TArk元舟

TArk Yuanzhou centers on model aggregation and a unified interface, catering to developers and teams that need to integrate multiple models. The official website offers a model catalog, an online chat feature, and an API key portal, allowing users to select supported models under a single service endpoint.

The official onboarding process includes registration, topping up your account as needed, and obtaining an API key, followed by configuring the service using an OpenAI-compatible SDK or client. The platform also outlines alternative channel routing and data review capabilities, but specific models, channels, and data processing conditions should be verified before integration.

Charges are based on the selected model and the content of each request; long contexts and longer outputs will affect actual usage. The latency, availability, and number of models displayed on the homepage are platform specifications and do not constitute a fixed guarantee for every request or all models, nor does the free‑start option equate to continuous free usage.

AI application developers and enterprise R&D teams

The official website states that recharge amounts are on demand, with actual consumption varying based on the model, context, and output length. The specific unit price per model has not been verified, so no uniform fixed plans are displayed, nor is a free access point presented as permanently free.