| Agent Workspace | Managed cloud workspace for your OpenClaw or Hermes agent. | An ongoing agent in ChatGPT, with its own cloud computer. |
|---|
| Model Providers | Choose supported OpenAI, Anthropic, Google, and other providers. | Uses OpenAI models; switching to third-party providers is not documented. |
|---|
| Your Own API Keys | Add keys for supported providers; provider usage is billed separately. | Bringing your own model API keys is not documented. ChatGPT plan access applies. |
|---|
| Agent Software | Choose OpenClaw or Hermes; MyClaw handles hosting and updates. | OpenAI manages the runtime; OpenClaw and Hermes are not documented options. |
|---|
| Background Work | Run ongoing tasks and scheduled workflows in your hosted workspace. | Continue work between conversations, with follow-ups and background task delegation. |
|---|
| Tools & Skills | Add supported integrations and reusable skills for your chosen runtime. | Use permitted plugins and cloud tools; local skills require a connected computer. |
|---|
| Messaging Channels | Web access plus configured Telegram, WhatsApp, Slack, Discord, and other supported channels. | ChatGPT messages and calls, Slack, and Teams. iMessage and Android RCS texting is a U.S. Pro beta. |
|---|
| What You Pay For | Hosting subscription plus model and third-party usage. AI tokens are not included in hosting. | Eligible ChatGPT plan and work allowances. Delegated Work or Codex tasks use those products’ limits. |
|---|
| Best Fit | You want to configure your agent’s providers, runtime, and integrations. | You want ongoing assistance connected to ChatGPT context and OpenAI’s tools. |
|---|