Skip to main content
OpenCode is a terminal-based coding agent that supports custom model providers. Add Tokios as a provider in opencode.json using the @ai-sdk/openai-compatible package, point options.baseURL at https://api.tokios.com/v1, and list your Tokios deployment name under models. OpenCode then routes every agent request down your private connector tunnel to your local model.

​
Prerequisites

  • A running Tokios connector paired with your local model — see Connector install
  • A registered model deployment name — see Register a model
  • A Tokios API key (sk-tok-…) — see API keys
  • OpenCode installed

​
Configure OpenCode

Add a Tokios provider block to your opencode.json config. The exact config file location (project-local opencode.json vs. a global config path) and the precise key names (options, npm, models) can vary across OpenCode releases, so confirm them against your installed version’s provider schema.
opencode.json
Replace gemma-tunnel with the public deployment name you registered on the Models tab — not the local model id your backend (Ollama, llama.cpp, vLLM, or LM Studio) uses. Replace sk-tok-YOUR_KEY with your own key, and avoid committing it into a shared or version-controlled config file.
options.baseURL must include /v1. OpenCode’s OpenAI-compatible provider sends requests relative to this URL, so a base URL without /v1 fails to reach the Tokios API surface.

​
Pick a tool-capable model

OpenCode’s agent relies on tool calling to read files, run commands, and edit code. Not every local model handles tool calls reliably — pick a deployment backed by a model with strong tool-calling support before you rely on OpenCode’s agent for real edits. See Choose a local model for how to compare candidates.

​
Troubleshooting

The API key OpenCode sent is missing, malformed, or has been revoked. Confirm options.apiKey holds the full sk-tok-… key with no extra whitespace, and check its status on the Keys tab.
The model key under models doesn’t match a registered deployment. Confirm it exactly matches the public name on the Models tab — not your Ollama, llama.cpp, vLLM, or LM Studio local model id.
The connector for that deployment is offline or at its concurrency cap. Check the Connectors tab — the connector needs to show as Online, and the machine running it needs to still be up and reachable.

​
Next steps

Compatible clients

See base URL and key setup for other OpenAI- and Anthropic-compatible clients.

Choose a local model

Compare backends and models before you commit a deployment to agentic workflows.