Editors and terminal agents are moving toward "bring your own model". If your model runs on a machine at home, the only missing piece is a URL the tool can reach.

Three values, every tool

Whatever the tool, you need the same three facts:

  • Base URL. https://tunnel.tokmine.ai/t/<slug>, with /v1 added for OpenAI-style clients.
  • Token. Your own account token from the portal.
  • Model name. Exactly as the model list returns it, for example llama3.1:8b.

Verify before you configure

Call the model list with your token first. If that works, any later failure is a client setting, not the network.

curl https://tunnel.tokmine.ai/t/<slug>/v1/models \
  -H "Authorization: Bearer $TOKMINE_TOKEN"

Use a static endpoint for editor configs

Anonymous session URLs expire after 30 minutes, which is no good in a config file. A static endpoint keeps the same URL, so you set it once.

Tool notes

Claude Code speaks the Anthropic Messages API, so use the URL without /v1 and a recent Ollama. Anthropic does not support routing Claude Code to non-Claude models, so treat that setup as best effort. Codex needs a runner that supports the OpenAI Responses API. Aider, Continue, Cline and Zed take the OpenAI-style URL with /v1.

Full guide

Copy-paste configs for each tool are in AI tools and clients.