最后更新 09-16-2026
分类:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
TokenHub
TokenHub is an LLM API gateway and model marketplace that sends GPT, Claude, Gemini, DeepSeek, and many other models through one OpenAI-compatible endpoint and one prepaid workspace balance. Product and platform teams use it when they want vendor choice without juggling separate API formats, auth flows, and billing consoles for every provider.
Where many gateways focus on chat tokens alone, TokenHub groups text, image, video, and speech routes in the same catalog, publishes per-model input and output rates on its site, and documents how to point existing tools like Claude Code or Cline at its base URL. Swapping models is designed to be a change to the model name in an OpenAI SDK client pointed at https://us-api.tokenhub.com/v1, not a full client rewrite.
The about page targets individual developers, B2B SaaS companies, enterprise IT groups, and consultancies shipping AI for clients. Docs cover chat, legacy completions, generation tasks, and native Claude-format POST routes, which matters when a team outgrows a single OpenAI shim.
Workspaces fund shared usage in amounts such as $20, $100, or $500, then issue gateway keys with usage tracking, budget caps, and routing controls so finance and engineering share one view of spend across models.
OpenAI SDK 客户端使用基础 URL https://us-api.tokenhub.com/v1
模型页面列出输入、输出和缓存读取速率，供你提交流量前参考
工作空间充值显示在 $20、$100 和 $500 的共享余额中
应用指南涵盖 Claude Code、Cline、Roo Code 和 Open WebUI 设置路径
REST 文档包括聊天、完成、生成任务和 Claude 原生 POST 路由
网关路由在关于页面宣传重试和提供商备用机制
一个兼容 OpenAI 的界面可以连接多个模型系列，无需单独的供应商 SDK
公共模型目录在你路由生产流量前比较令牌和生成价格
已发布的应用集成指南减少了对 Claude Code、Cline 和类似 BYOK 工具的猜测
共享工作空间余额简化了多个团队使用不同模型时的财务追踪
预付余额计费意味着没有为试验提供文档化的免费层
每个模型的费率和可用性可能变化，因此生产应用需要持续检查目录
一些客户端应用需要兼容 Anthropic 的基础模型，而另一些则需要 OpenAI /v1，这增加了设置复杂性
Does TokenHub offer a free plan?
No. TokenHub bills through prepaid workspace balance instead of complimentary credits. You add funds such as , , or , then pay per model usage from that balance according to rates on TokenHub model catalog pages.
How much does TokenHub cost?
TokenHub charges per model and request type from your workspace balance. Its homepage highlights $20, $100, and $500 top-ups, while the models catalog lists per-million-token input and output prices for each model, such as DeepSeek V4.1 Flash at $0.2857 input per million tokens.
Is TokenHub compatible with the OpenAI API?
Yes. TokenHub documents an OpenAI-compatible REST API and shows sample OpenAI SDK code using base URL https://us-api.tokenhub.com/v1. You keep familiar chat completion request patterns after updating the base URL and API key.
What models can I call through TokenHub?
TokenHub maintains a browsable models catalog with text, image, video, and speech routes from providers such as OpenAI, Google, Anthropic, DeepSeek, Qwen, Moonshot AI, MiniMax, and Tencent Hunyuan. Each entry lists modalities, context limits, and listed token rates.
How do I connect Claude Code to TokenHub?
TokenHub apps section includes a Claude Code setup guide that uses an Anthropic-compatible API base, your TokenHub API key, and a supported model name. Anthropic-format apps use the Messages API path documented in TokenHub integration guides rather than the OpenAI /v1 base alone.
What do I need to start using TokenHub?
TokenHub homepage workflow is create a workspace with email or Google sign-in, fund shared usage, then issue a gateway API key in the console. Most integrations need that key plus the documented API base and a model name from the catalog.
Does TokenHub support coding agents and chat UIs?
Yes. TokenHub publishes setup guides for tools including Cline, Roo Code, OpenHands, LangChain, Open WebUI, Cursor, and Zed Editor. Each guide notes whether the app expects an OpenAI-compatible or Anthropic-compatible base URL before you enter your TokenHub key.