NextHub FAQ
Check here first when you run into a problem.
What should I do about a 429 error?
429 means you have exceeded your concurrency limit or your balance is insufficient. Check that your account quota and balance are sufficient; if your quota is sufficient and you still trigger 429 frequently, you may have hit the rate limit of your ApiKey or group (5-hour / daily / weekly windows). Try again later or adjust your rate limit settings.
What about "Model unavailable" or "Request rejected"?
It is most likely a group or model ID problem. The group you select when creating an ApiKey determines which models you can call, so confirm that the target model belongs to the selected group and that the model ID is written exactly as in the model list.
Why can the claude command still not connect after saving settings.json?
Confirm the file path is correct (Windows: %USERPROFILE%\.claude\settings.json; macOS: ~/.claude/settings.json), check that the JSON is valid (no trailing comma on the last line), and verify that ANTHROPIC_BASE_URL is https://api.nexthub.cc (do not append /v1).
What groups are available, and what are their multipliers?
There are currently 9 public groups (snapshot dated 2026-09-27): claude code (0.37x, no client restrictions), Claude Max (0.5x, high stability and high cache hit rate), Claude-AWSB (0.7x), GPT-Azure (0.47x), GPT-Pro (0.11x, no image generation), Gemini-aistudio (0.28x), Gemini-Vertex (0.49x), Grok (0.55x), 国产模型组 (0.15x). Dedicated groups such as image generation are also available and become visible after you log in to the console and enable the corresponding permissions. Actual charge = official price × group multiplier; see Group Multipliers for details.
Claude Desktop shows a warning after configuration and never shows a "Test passed" message?
This is normal. Claude Desktop's third-party gateway mode does not go through official login verification, so the configuration page never shows a "Connected" message, and the "Configuration unverified" / "Network connection" warning bar on the chat page after a restart is also a routine false alarm. Just send a message: if you receive a normal reply, the configuration is working. See the Claude Desktop Configuration Guide for details.
What should I do if Zed reports request errors after configuration?
First confirm that the Base URL is https://api.nexthub.cc/v1 (with /v1). This site supports both the chat/completions and Responses request modes, so you can keep the default; the model ID must match the model list (for example claude-opus-5-5).
Why does Kilocode give no reply after connecting?
Check that the Base URL is filled in correctly (https://api.nexthub.cc/v1, with /v1) and confirm that the API Provider is set to OpenAI Compatible (Responses) rather than any other type.