Skip to main content

​
Overview

Type: Coding ToolPrimary Path: OpenAI-compatibleSupport Confidence: Supported path
OpenCode is an open-source AI coding assistant that runs in your terminal. It supports multiple LLM providers and can be configured to use AI Sonar’s API for access to hundreds of models.

​
Installation

curl -fsSL https://opencode.ai/install | bash
Verify installation: 
opencode --version

​
Configuration

​
Step 1: Set Environment Variables

export OPENAI_API_KEY="sk-your-api-key"
For permanent configuration, add to ~/.bashrc or ~/.zshrc: 
echo 'export OPENAI_API_KEY="sk-your-api-key"' >> ~/.zshrc
source ~/.zshrc

​
Step 2: Configure OpenCode

OpenCode’s current docs recommend configuring providers through opencode.json or opencode.jsonc. For AI Sonar, use a custom provider instead of overloading the built-in openai provider: 
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "aisonar": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "AI Sonar",
      "options": {
        "baseURL": "https://api.aisonar.dev/v1",
        "apiKey": "{env:OPENAI_API_KEY}"
      },
      "models": {
        "gpt-5.4": {
          "name": "GPT-5.4"
        },
        "gpt-5-mini": {
          "name": "GPT-5 Mini"
        }
      }
    }
  },
  "model": "aisonar/gpt-5.4"
}
Use @ai-sdk/openai-compatible for chat-completions style compatibility. If you specifically need a provider path built on /v1/responses, OpenCode’s provider docs recommend switching the package to @ai-sdk/openai.

​
Basic Usage

Start interactive mode: 
opencode
Run with a prompt: 
opencode "Explain this codebase"
Specify model: 
opencode --model aisonar/gpt-5.4 "Fix the bugs in main.py"

​
Available Models

ModelBest For
gpt-5.4Complex tasks, code architecture
gpt-5-miniQuick fixes, simple queries
claude-sonnet-4-6Code review, documentation
claude-opus-4-6Complex reasoning
gemini-2.5-flashFast responses
deepseek-r1Algorithm design

​
Common Commands

Analyze code: 
opencode "What does this function do?" < src/utils.ts
Generate code: 
opencode "Create a REST API with Express"
Review changes: 
git diff | opencode "Review these changes"
Fix errors: 
opencode "Fix the TypeScript errors in this project"

​
Interactive Commands

CommandDescription
/helpShow available commands
/model <name>Switch to a different model
/clearClear conversation history
/exitExit OpenCode

​
Troubleshooting

  • Verify options.baseURL is set to https://api.aisonar.dev/v1
  • Check network connectivity
  • Try curl https://api.aisonar.dev/v1/models to test
  • Verify OPENAI_API_KEY environment variable is set
  • Verify options.apiKey references {env:OPENAI_API_KEY} or another valid secret source
  • Check that the key starts with sk-
  • Ensure the key is active in AI Sonar dashboard

​
Best Practices

Run OpenCode from your project root for better understanding of your codebase.
Use faster models (gpt-5-mini) for simple tasks and stronger models (gpt-5.4, claude-sonnet-4-6) for complex ones.
Always review AI-generated code before applying changes to your project.