> ## Documentation Index
> Fetch the complete documentation index at: https://oxenai-eric-document-openclaw.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenClaw

> Use Oxen.ai models from OpenClaw, a local personal AI assistant.

[OpenClaw](https://github.com/openclaw/openclaw) is a personal AI assistant you run on your own machine. It connects to chat apps like WhatsApp, Telegram, Discord, and Slack, and supports custom OpenAI-compatible model providers. This guide points OpenClaw at Oxen.ai so every agent turn runs through the Oxen [chat completions API](/examples/inference/chat_completions).

## Prerequisites

* Node 24 (recommended) or Node 22.19+
* An [Oxen.ai API key](https://www.oxen.ai/settings/profile), exported as `OXEN_API_KEY`

## Install OpenClaw

```bash theme={null}
npm install -g openclaw@latest
```

## Add Oxen as a model provider

Run `openclaw onboard` with the `custom-api-key` flow:

```bash theme={null}
openclaw onboard \
  --non-interactive \
  --accept-risk \
  --auth-choice custom-api-key \
  --custom-provider-id oxen \
  --custom-base-url https://hub.oxen.ai/api/ai \
  --custom-api-key "$OXEN_API_KEY" \
  --custom-model-id claude-sonnet-4-6 \
  --custom-compatibility openai \
  --custom-text-input \
  --install-daemon
```

Output:

```
Updated config: ~/.openclaw/openclaw.json
Workspace OK: ~/.openclaw/workspace
Sessions OK: ~/.openclaw/agents/main/sessions
```

`--custom-model-id` sets the default model. Pick any text model from the [Oxen catalog](https://www.oxen.ai/ai/models); a capable general-purpose model like `claude-sonnet-4-6` is a good starting choice. `--install-daemon` installs the gateway as a background service so it stays running.

This writes a `models.providers.oxen` block to `~/.openclaw/openclaw.json` with `baseUrl`, `apiKey`, `api: "openai-completions"`, and a one-entry `models` list.

## Run an agent turn

```bash theme={null}
openclaw agent --agent main --message "What is a great name for an ox? Reply in 5 words or less."
```

Output:

```
**Bruno** works perfectly for an ox.
```

The agent calls `https://hub.oxen.ai/api/ai/chat/completions` under the hood, billed to your Oxen account.

## Add more models

To expose a second Oxen model, edit `models.providers.oxen.models` in `~/.openclaw/openclaw.json` and append an entry with the same shape:

```json theme={null}
{
  "id": "gpt-5-mini",
  "name": "GPT-5 Mini",
  "input": ["text"],
  "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
  "contextWindow": 200000,
  "maxTokens": 4096,
  "reasoning": false
}
```

Then switch the default:

```bash theme={null}
openclaw models set oxen/gpt-5-mini
```

## Next steps

* Connect OpenClaw to [WhatsApp](https://docs.openclaw.ai/channels/whatsapp), [Telegram](https://docs.openclaw.ai/channels/telegram), [Discord](https://docs.openclaw.ai/channels/discord), or any other [supported channel](https://docs.openclaw.ai/channels) so you can chat with your Oxen-backed assistant from anywhere.
* Browse the full [Oxen model catalog](https://www.oxen.ai/ai/models). Vision models work through the same provider; see [Vision Language Models](/examples/inference/vision_language_models) for the message shape.