OneLLM · Reviewed November 12, 2026
Use DeepSeek Harness from your iPhone or iPad.
OneLLM connects your phone to DeepSeek Harness running on your computer. Open the DS Harness module and scan that service’s connection code. The computer does the work and must stay on and reachable. This is an independent client, not an official DeepSeek app.
Choose the connection that matches your work
- OneLLM Office: continue OneLLM Office tasks and conversations from the phone. Get the computer app from OneLLM Office downloads.
- DS Harness: connect to your existing DeepSeek Harness service using the DS Harness phone module.
- OpenCode: use the separate OpenCode module for OpenCode projects. Read the OpenCode mobile setup guide.
These connection codes are different. Installing OneLLM Office does not turn an OneLLM Office pairing code into a DS Harness or OpenCode code.
Before you pair
- Install OneLLM on your iPhone or iPad.
- Confirm DeepSeek Harness opens and can complete a short request on the computer.
- Configure its model account on the computer. Remote access does not include model usage or bypass that account’s limits.
- Keep both devices reachable over a trusted local network or Tailscale. A home LAN address normally cannot be reached from mobile data.
Pair and check the result
- Follow the computer connection instructions for your platform. Connection setup is integrated into OneLLM Office. Open Phone & computers to set up the connection.
- Select DS Harness, the intended workspace and the appropriate network address. Wait for the computer service to become ready.
- On your phone, open OneLLM → DS Harness → Scan connection code. Use a freshly generated DS Harness code from your own computer.
- Open the intended conversation. Send a short request and confirm a reply arrives before trying a larger task.
- If connecting away from home, check Tailscale on both devices and use a reachable Tailscale address.
If the connection does not work
| Symptom | What to check |
|---|---|
| The code is rejected | Check the selected module and generate a new DS Harness code. Do not scan an OneLLM Office or OpenCode code here. |
| It works on Wi-Fi but not away from home | Check whether you paired with a LAN-only address. Verify the private-network connection before changing model settings. |
| Connected, but no model response | Try the same request on the computer. Check its model configuration and usage limits; repeating pairing will not fix a model-account error. |
| An old conversation appears | Check the selected workspace and conversation on the computer. |
| The connection stops | Check computer sleep, service shutdown and network changes. Reconnect to a running service. |
Does this run DeepSeek Harness offline on my phone?
No. This is remote access to a computer service. OneLLM’s downloaded on-device chat models are a separate feature. They do not keep a remote service available when the computer is off.
Do I need to expose a public port?
Use a trusted LAN or a private-network connection such as Tailscale. Keep connection codes private because they may contain credentials. Follow your service’s authentication and firewall instructions.
Official documentation and next steps
Read the DeepSeek Harness repository for current service requirements. For a workflow comparison, see DeepSeek Harness vs OpenCode. Set up OneLLM Office phone access or explore OneLLM Office.