https://token.subfor.me/v1Join the beta
- Ask the administrator to create your account, then sign in. Public registration is currently closed.
- Ask your inviter to allocate test credits. New accounts start with zero credits.
- Create a budget-limited API key and set your client endpoint.
Available today
The first models are connected through CLIProxy. Use GET /v1/models for the models available to your account. Names are declared by the upstream and are not independent identity certifications.
The quickstart covers Chat Completions, streaming and tool calls. Codex++ one-click setup, automated quality monitoring and model fingerprint tools are not yet available.
GPT pricing
Default group prices in CNY per 1 million tokens, converted at USD 1 = CNY 7.3; checked 2026-10-30. See the TokenHub console for current rules.
| Model and context | Input | Output | Cache read | Cache write |
|---|---|---|---|---|
| GPT-6 Astra | ¥73 | ¥365 | ¥7.30 | ¥91.25 |
| GPT-6.1 Sol (≤272K) | ¥14.60 | ¥73 | ¥0.73 | ¥18.25 |
| GPT-6.1 Sol (>272K) | ¥29.20 | ¥109.50 | ¥1.46 | ¥36.50 |
Credits and pricing
Account balances and usage charges are displayed in Chinese yuan in the console. Input, output, and cached tokens are billed according to each model’s pricing rules; see the TokenHub console for current prices and individual charges. Online payments and top-ups are not enabled.
The ledger uses upstream-reported usage. Cache and reasoning token accounting depends on the protocol; inspect individual usage records in the console.
Data and beta scope
The service forwards your requests to configured model providers and records operational information such as account, model, time, usage and errors. Avoid sensitive or irreplaceable data during the beta.
The performance page publishes only fixed synthetic tests, never customer conversations, keys or account balances.