How to switch models on our own?

[BradKML]

@WitMiao feels like my config broke, it still stuck using Gemini, but I pretty much forced the whole router to use only GLM, so I am not sure what is happening for ~/.claude-code-router/config.json.

{
  "LOG": true,
  "CLAUDE_PATH": "",
  "HOST": "127.0.0.1",
  "PORT": 3456,
  "APIKEY": "sk-zcf-x-ccr",
  "API_TIMEOUT_MS": "600000",
  "PROXY_URL": "",
  "transformers": [],
  "Providers": [
    {
      "name": "openrouter",
      "api_base_url": "https://openrouter.ai/api/v1/chat/completions",
      "api_key": "sk-or-v1-dummy-text-whoops",
      "models": [
        "google/gemini-2.5-pro-preview",
        "anthropic/claude-sonnet-4",
        "anthropic/claude-3.5-sonnet",
        "anthropic/claude-3.7-sonnet:thinking",
        "z-ai/glm-4.5-air",
        "z-ai/glm-4.5-air:online"
      ],
      "transformer": {
        "use": [
          "openrouter"
        ]
      }
    }
  ],
  "Router": {
    "default": "openrouter,z-ai/glm-4.5-air",
    "background": "openrouter,z-ai/glm-4.5-air",
    "think": "openrouter,z-ai/glm-4.5-air",
    "longContext": "openrouter,z-ai/glm-4.5-air",
    "longContextThreshold": 60000,
    "webSearch": "openrouter,z-ai/glm-4.5-air:online"
  }
}

If I delete the lines for Gemini and Claude in "models", and then claude:

  ⎿  API Error (401 {"error":{"message":"Error from provider(openrouter,google/gemini-2.5-pro-preview: 401): {\"error\":{\"message\":\"User not found.\",\"code\":401}}Error: Error from provider(openrouter,google/gemini-2.5-pro-preview: 401): {\"error\":{\"message\":\"User not found.\",\"code\":401}}\n    at nt (/home/user/.npm-global/lib/node_modules/@musistudio/claude-code-router/dist/cli.js:75644:11)\n    at u0 (/home/user/.npm-global/lib/node_modules/@musistudio/claude-code-router/dist/cli.js:75702:11)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at async o0 (/home/user/.npm-global/lib/node_modules/@musistudio/claude-code-router/dist/cli.js:75669:84)","type":"api_error","code":"provider_response_error"}}) · Retrying in 1 seconds… (attempt 1/10)

Gemini is like ~10x more expensive than open weight models, so I kinda want an easy fix for this

[WitMiao]

After modifying the configuration, use ccr restart.
I just used the configuration you provided, and it showed: API Error (401 {"error":{"message":"Error from provider(openrouter,z-ai/glm-4.5-air: 401). The explanation is that the glm4.5-air was called normally, it's just that I don't have a token.

[BradKML]

Will need to try that later on

[BradKML]

Welp that is interesting #39

[WitMiao]

There is a chance that the ccr service may freeze. You can use the following commands to terminate it and then restart it.

# windows
# 1. First, get the PID of the service on port 3456 like this
netstat -ano | findstr :3456
# Get the following content
# TCP    127.0.0.1:3456         0.0.0.0:0              LISTENING       1208
# TCP    127.0.0.1:59047        127.0.0.1:3456         TIME_WAIT       0
# 2. Then terminate it. 1208 is not a fixed value; it depends on the PID obtained from the previous command.
taskkill //PID 1208 //F

#mac/linux Execute the following command directly
lsof -t -i:3456 | xargs kill

[BradKML]

Would love to see documentation covering this trouble-shooting process

[WitMiao]

Working on it, working on it.

The two things programmers dislike the most🤣:

1. Why there’s no documentation
2. Not wanting to write documentation