CodeBuddy Custom Models: Claude & GPT

CodeBuddy is Tencent Cloud's AI code editor; its models.json accepts full OpenAI-compatible endpoints, so one Code By AI key serves Claude and GPT.

CodeBuddy reads custom models from ~/.codebuddy/models.json (user level) or <workspace>/.codebuddy/models.json (project level). Project entries override user entries with the same id; the file hot-reloads about a second after you save.

One command (macOS / Linux)

Copy
1
curl -fsSL https://www.codebyai.net/setup.sh | sh -s -- --codebuddy

It asks for your API key (hidden input), adds gpt-5.6-terra and claude-sonnet-5 to the user-level ~/.codebuddy/models.json, backs up the file, and leaves it owner-only. CodeBuddy reloads the file by itself; pick either model in the model picker. Re-running rotates the key; | sh -s -- --uninstall restores the file. A project-level .codebuddy/models.json that defines the same id still wins over these entries. The CodeBuddy plugin does not read models.json; use the IDE or CLI. The manual steps below do the same by hand.

1. Create a Code By AI API key

Sign in to Code By AI, open API Keys, and create a key. It looks like sk-proj-....

2. Edit models.json

Open ~/.codebuddy/models.json and merge this into the existing object. Keep any models already there:

Copy
1
{
2
  "models": [
3
    {
4
      "id": "gpt-5.6-terra",
5
      "name": "CodeByAI GPT",
6
      "vendor": "Custom",
7
      "url": "https://www.codebyai.net/api/agent/v1/chat/completions",
8
      "apiKey": "sk-proj-xxxx",
9
      "maxInputTokens": 200000,
10
      "maxOutputTokens": 65536,
11
      "supportsToolCall": true,
12
      "supportsImages": false,
13
      "supportsReasoning": false
14
    },
15
    {
16
      "id": "claude-sonnet-5",
17
      "name": "CodeByAI Claude",
18
      "vendor": "Custom",
19
      "url": "https://www.codebyai.net/api/agent/v1/chat/completions",
20
      "apiKey": "sk-proj-xxxx",
21
      "maxInputTokens": 200000,
22
      "maxOutputTokens": 65536,
23
      "supportsToolCall": true,
24
      "supportsImages": false,
25
      "supportsReasoning": false
26
    }
27
  ]
28
}

CodeBuddy only speaks the OpenAI interface format, so url must be the complete Chat Completions path. Claude IDs still work: the gateway forwards claude-* to the Claude channel and bills Claude prices.

3. Select the model

Save the file. CodeBuddy reloads it automatically; open the model picker and choose CodeByAI GPT or CodeByAI Claude. If you maintain availableModels, add both IDs there, otherwise the list is ignored and every configured model is shown.

Model fields

• maxInputTokens and maxOutputTokens should match the model's real limits; 200,000 and 65,536 are the values the gateway's models use.
• supportsToolCall must stay true for agent tasks.
• supportsImages and supportsReasoning are optional; enable them only when the selected model supports image input or reasoning parameters.

Troubleshooting

• 404 — url is a base URL or ends somewhere other than /chat/completions. Use the full https://www.codebyai.net/api/agent/v1/chat/completions.
• 401 / invalid key — the key has stray spaces, was deleted, or expired. Create a new one on the API Keys page.
• Insufficient balance — top up on the Wallet page.
• The model is hidden — a non-empty availableModels list does not include its ID. Add it or remove the list.
• The file is ignored — invalid JSON stops the reload. CodeBuddy's file is plain JSON; remove comments and trailing commas.