OpenCode
Prepare API Access
Have the three items below ready before you configure anything:
API Key
The key string you generated when creating your API Key.
Base URL
https://api.watt-cloud.com/v1
The trailing /v1 is required — do not drop it.
Model Name
Take the complete model identifier from the Model Marketplace, for example DeepSeek-V4-Pro.
Installing OpenCode
Install OpenCode through one of its official channels; the OpenCode documentation covers the details.
Linux / macOS / WSL2
curl -fsSL https://opencode.ai/install | bash
Prefer npm? Install globally:
npm install -g opencode-ai
Windows
WSL2 is OpenCode's own recommendation for the smoothest Windows experience; run the Linux install command above from inside your WSL2 environment.
On native Windows, use Chocolatey or Scoop instead:
choco install opencode
# or
scoop install opencode
npm works here as well:
npm install -g opencode-ai
Connect OpenCode to Watt Cloud
A custom provider entry in ~/.config/opencode/opencode.json is what OpenCode uses to route its requests.
Edit the Configuration File
Open ~/.config/opencode/opencode.json — create the file first if it doesn't exist — and add:
{
"$schema": "https://opencode.ai/config.json",
"model": "watt/GLM-5.1",
"provider": {
"watt": {
"npm": "@ai-sdk/openai-compatible",
"name": "Watt Cloud",
"options": {
"baseURL": "https://api.watt-cloud.com/v1",
"apiKey": "{env:WATT_API_KEY}"
},
"models": {
"GLM-5.1": {
"name": "GLM-5.1"
},
"DeepSeek-V4-Pro": {
"name": "DeepSeek-V4-Pro"
},
"MiniMax-M2.7": {
"name": "MiniMax-M2.7"
}
}
}
}
}
Field Reference
| Field | Description |
|---|---|
model | Model OpenCode uses by default, written as provider/model (e.g., watt/GLM-5.1) |
npm | The AI SDK package to load; pick @ai-sdk/openai-compatible for /v1/chat/completions |
name | Human-readable label shown for this provider |
options.baseURL | Endpoint of the Watt Cloud API; the trailing /v1 is required |
options.apiKey | Your key — pass {env:VAR_NAME} to pull it from the matching environment variable |
models | The model IDs, exactly as the Model Marketplace lists them |
Note: The top-level model entry is mandatory. Every ID you list under models must match a Model Marketplace name exactly.
Set the API Key
Export the key before launching OpenCode:
export WATT_API_KEY="your-api-key"
To keep the variable across sessions, put that export line into a shell profile such as ~/.bashrc or ~/.zshrc.
Alternatively, opencode providers login handles this interactively: choose Other, type the provider ID watt, then paste in your key.
Change the Model
Two ways to do it: edit the top-level model entry in the config, or supply -m as a CLI flag:
"model": "watt/DeepSeek-V4-Pro"
Verify the Connection
Once WATT_API_KEY is exported, run:
opencode run -m watt/GLM-5.1 "Reply with exactly one word: hello"
If everything is wired up correctly, the response comes back as pong.
Launch the TUI
Start OpenCode's interactive TUI from your project's directory:
opencode
Inside the TUI, /models lets you switch between models.