xmr4ai

opencode integration

Terminal coding agent. Talks to any OpenAI-compatible provider.

Create a key
Mint one on the API keys page and name it opencode, so it can be revoked on its own if the machine it lives on is lost.
2
Put the key in your environment
The config below reads it from there rather than holding the secret itself, so the file stays safe to commit to a repo.
```
export XMR4AI_API_KEY="xmr4ai_live_..."
```

Register the provider

Add this to ~/.config/opencode/opencode.json for every project, or to opencode.json in a repo to scope it there. Models are declared explicitly — opencode cannot enumerate a custom provider’s catalogue, so anything you want to select has to be listed.

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "xmr4ai": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "xmr4ai",
      "options": {
        "baseURL": "https://xmr4ai.com/api/v1",
        "apiKey": "{env:XMR4AI_API_KEY}"
      },
      "models": {
        "anthropic/claude-sonnet-4": {
          "name": "Claude Sonnet 4",
          "limit": { "context": 200000, "output": 64000 }
        },
        "openai/gpt-4.1-mini": {
          "name": "GPT-4.1 Mini",
          "limit": { "context": 1000000, "output": 32000 }
        }
      }
    }
  }
}

Or hand the key to opencode directly
Instead of the environment variable, run /connect inside opencode, choose Other, enter the provider id xmr4ai, and paste the key. It is stored in opencode’s own credentials file rather than in the config you just edited — drop the apiKey line if you go this route.
5
Run it
The model argument is the provider id, then the model id, separated by a slash. /models picks one interactively instead.
```
opencode run -m xmr4ai/anthropic/claude-sonnet-4 "explain this repo"
```

opencode integration

Create a key

Put the key in your environment

Register the provider

Or hand the key to opencode directly

Run it