AI Provider Setup Guide

OpenCode Configuration

OpenCode is an open-source terminal AI coding assistant. It's suitable for code Q&A, file modifications, and model switching within a project directory. Beginners should first use its own connection wizard and avoid manually writing the complete JSON at the start.

Official information Downloads and documentation are subject to the official site.

OpenCode Official Website Open-source terminal AI coding assistants Documentation Installation, authentication, and command instructions AI provider configuration reference Refer to the configuration idea in opencode.json

What this tool is and what it is used for

Configuration File Location

system	Configuration File Location	Description
Windows	`C: \Users\YourUsername\. config\opencode\opencode.json`	If the file doesn't exist, create a new one
macOS / Linux	`~/. config/opencode/opencode.json`	If the file doesn't exist, create a new one

Start with the connection wizard

OpenCode Connection Wizard

cd YOUR_TEST_PROJECT opencode /connect

OpenCode Check Model

cd YOUR_TEST_PROJECT opencode /models

What to pay attention to when configuring manually

Only edit the configuration file if your provider's documentation clearly specifies the OpenCode Provider field, where to place the Base URL, and Key. Fields may vary between versions, do not directly copy JSON examples from other tools into OpenCode.

Items to verify	Description
Provider Type	Do not mix OpenAI, Claude, Gemini, or other protocols
Base URL	Use the complete address provided by the provider for OpenCode or the corresponding protocol
Key	Use the Key corresponding to the current Provider
Model Name	Copy the model list from the provider's backend

How to test after saving

Run after entering a test project directory opencode.
First use /models Confirm that the model you just configured is visible.
After selecting the model, send 'Please only reply: OpenCode configuration successful'.
The first time, do not let it modify files; just let it explain the current directory or the current file.
If you fail, go back first /connect Or check the provider, base URL, key, and model name against the provider's documentation.

Phenomenon	Priority check	Handle
`/models` There is no model inside	Whether the provider saved successfully, whether the model list can be retrieved	Manually add provider backend model ID / model name
Connection failed	Whether the Base URL belongs to the current Provider protocol	Do not mix addresses of Claude, Gemini, and OpenAI
The configuration file is invalid	JSON format, commas, quotation marks	Prioritize generating the configuration using the connection wizard, then make minor manual adjustments.

OpenCode Official Website View installation, authentication, and command instructions.

OpenCode AI provider Configuration Reference Refer to the configuration approach of opencode.json.