Models
Configure LLM providers and models in xCloud.
xCloud uses OpenClaw model providers. The model selector lets you search providers, choose a model, and assign it to the current agent.
Model Scope
Models are selected per agent.
That means your coding agent can use one model, your research agent can use another, and your workspace coordinator can use a third.
Provider Credentials
Some providers use API keys. Others use account login profiles.
| Provider type | Examples | Auth style |
|---|---|---|
| API key | Anthropic, OpenAI API, Google, Groq, Mistral, DeepSeek | Paste key in Settings > API Keys |
| Account auth | OpenAI Codex, GitHub Copilot | Login/auth profile |
| Local | Ollama | Local server, usually no key |
Read API Keys before configuring OpenAI Codex or GitHub Copilot.
OpenAI API vs OpenAI Codex
This is the most common confusion.
| Choice | What it means |
|---|---|
| OpenAI API | Uses an API key from the OpenAI platform |
| OpenAI Codex | Uses Codex/ChatGPT account access through an auth profile |
Connecting an OpenAI API key does not automatically authenticate OpenAI Codex models.
Choose A Model
- Open an agent.
- Open the model selector from the chat input.
- Search for a provider or model.
- Select the model.
- Send a short test message.
If the provider needs credentials, configure them first in Settings > API Keys.
Remote Engines
Model requests are made by the active engine.
If the active engine is a VPS, the VPS needs the provider credentials. A key saved on Local will not automatically exist on the VPS.
Ollama
For local models:
- Install Ollama.
- Pull a model, for example
ollama pull llama3.2. - Make sure the active engine can reach
http://localhost:11434. - Select an Ollama model in xCloud.
If the engine is remote, localhost means the remote machine, not your Mac.