Frequently asked questions
Common questions about HopBase integration, keys, protocols, models, and clients.
No. HopBase is directly accessible from mainland China without a proxy.
OpenAI protocols for GPT, Codex, and third-party clients include /v1. The Anthropic protocol for Claude Code and the Anthropic SDK does not; those clients append the API path themselves.
Select OpenAI Chat Completions (routing required), use https://api.hop-base.com/v1, and set the model to glm-5.2.
Use the OpenAI Images protocol: POST https://api.hop-base.com/v1/images/generations with model, prompt, and size. Gemini Banana and Seedream 5.0 Pro use the same client-side protocol, but image-to-image and asynchronous capabilities vary. Seedream supports synchronous generation and editing with up to 10 reference images. Add image to the generations JSON or use /v1/images/edits. See the Image API.
No. Each key belongs to one plan group. Use an OpenAI plan key for GPT and a Claude plan key for Claude. Create both in the console when your application needs both families.
Yes. The console's one-click integration dialog provides a Windows PowerShell command. Each manual setup page also includes Windows instructions.
The command contains a single-use integration token that expires after 15 minutes, not the API key itself. The script only updates local configuration and backs up existing settings.
Yes. Usage from the same key is combined under one account. Separate named keys per device make usage easier to attribute.
Sign in to the console. Usage Statistics shows the model, token count, and charge for every request.
This usually indicates temporary instability for that model. Retry shortly and contact us if it persists. See Troubleshooting for other errors.
Direct browser connections are supported. Web clients often do not append an API path. For OpenAI, GPT, or GLM modes, enter the complete https://api.hop-base.com/v1 URL. For Anthropic mode and Claude, omit /v1.