Use BigModel.io with OpenCode

OpenCode supports any OpenAI-compatible API through the @ai-sdk/openai-compatible provider. It calls /v1/chat/completions, so it works with every BigModel.io group that serves text models, including Chat Completions-only GLM and DeepSeek groups.

#Configure opencode.json

Put this in ~/.config/opencode/opencode.json (all projects) or opencode.json in a project root:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "bigmodel": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "BigModel.io",
      "options": {
        "baseURL": "https://bigmodel.io/v1",
        "apiKey": "{env:BIGMODEL_API_KEY}"
      },
      "models": {
        "glm-5.3": { "name": "GLM-5.3" },
        "glm-5.3-flash": { "name": "GLM-5.3 Flash" },
        "kimi-k3": { "name": "Kimi K3" },
        "deepseek-v4-pro": { "name": "DeepSeek V4 Pro" }
      }
    }
  },
  "model": "bigmodel/glm-5.3",
  "small_model": "bigmodel/glm-5.3-flash"
}

Each key under models must be the exact model ID; name is only the label OpenCode shows. {env:BIGMODEL_API_KEY} reads the key from your environment, so the file is safe to share. model and small_model use the provider/model form; small_model handles light tasks such as titles.

export BIGMODEL_API_KEY="sk-..."
opencode

#Set context limits

OpenCode does not know the context window of a custom model. Add limit to each model so it compacts history at the right point. Take the numbers from the model page; the values below are placeholders:

"glm-5.3": {
  "name": "GLM-5.3",
  "limit": { "context": 128000, "output": 16000 }
}

#Store the key with /connect instead

If you prefer not to use an environment variable, run /connect in OpenCode, choose Other, enter the provider ID bigmodel and paste your key. OpenCode stores it in ~/.local/share/opencode/auth.json. Then remove apiKey from options.

#Verify the setup

Run /models inside OpenCode. The BigModel.io models from your config should be listed. Pick one, send a short prompt, and check that the request appears in Usage.

#Troubleshooting

The BigModel.io provider does not appear in /models. The JSON is invalid or in the wrong place. Validate it, and check that a project opencode.json is not replacing your global file.

401 API_KEY_REQUIRED. BIGMODEL_API_KEY is empty in the shell that started OpenCode, or the provider ID used with /connect does not match the key under provider.

503 api_error: No available accounts: no available accounts supporting model: .... The key under models does not match a model your key's group serves. Copy the ID from the model page.

404 on every request. baseURL is missing /v1 or has an extra /chat/completions.

Long sessions fail or get cut off. limit.context is unset or too large. Set it to the model's real context window.

Tool calls fail or loop. Some models handle long agent sessions better than others. Try a different model, and keep the first task small.

403 INSUFFICIENT_BALANCE or 429. Top up in Billing. A 429 rate_limit_error ("Too many pending requests") means too many parallel requests; wait and retry.

#Related

• Configure Base URL in Codex, Claude Code, Cursor and Cline
• Claude Code and Codex CLI, the other terminal agents.
• Migrate from OpenRouter if your config uses provider/model names.