How to connect Claude Code to Qwen and GLM, step by step

How it works

Claude Code sends requests to the address in ANTHROPIC_BASE_URL. Point it at Tenzo Cloud and the requests come to us while the chosen model answers. The interface, commands and agent features stay the same.

Setup

1. 1

   Create a key

   Sign up at Tenzo Cloud and create a key under API keys. The first 3 hours are free.

2. 2

   Set the variables

   In a terminal in your project folder, set the address, key and model.

   terminalCopy

   export ANTHROPIC_BASE_URL=https://tenzo-cloud.com
   export ANTHROPIC_AUTH_TOKEN=tc-your-key
   export ANTHROPIC_MODEL=glm-5.3
   export ANTHROPIC_SMALL_FAST_MODEL=glm-5.3
   claude

3. 3

   Make it permanent

   Add the same values to the env block of ~/.claude/settings.json so you do not set them in every terminal.

Which model to pick

See the Models page for each model’s current status.

If something does not work

Claude Code asks you to log in

The variables are not visible in the current terminal. Set them again or via settings.json and restart claude.

Address error

Claude Code needs the address without /v1: https://tenzo-cloud.com.

429 while the agent works

The agent runs parallel requests. For active work we recommend 4 streams.

Several projects, different settings

Environment variables can be set per project. If you want Tenzo Cloud in one repository and your previous account in another, put the settings in .claude/settings.json inside that project and leave the global ~/.claude/settings.json as is.

It is also handy to keep different models: GLM 5.3 for a backend with long agent tasks and Qwen 3.8 Max for a frontend where you often share screenshots.

Tips for long sessions

• Give the agent a project rules file so it asks fewer questions
• Take 4 streams for parallel work; Claude Code sometimes sends several requests at once
• Restrict the key by IP if you run the agent on a server
• Watch the status page; during maintenance switch models with one variable

Why this beats a capped subscription

Chat subscription limits run in time windows: use them up and you wait. With Tenzo Cloud only concurrent requests are limited, so the agent can work all day without pauses. The price is fixed for 3, 7 or 30 days, no token bill.