OpenCode

OpenCode is an open-source, terminal-based AI coding assistant. It runs in your terminal, understands your codebase, and can write, edit, and debug code across files.

OpenCode uses the OpenAI-compatible Chat Completions API, making it a straightforward integration with Token101.

Prerequisites

• A Token101 API key — subscription users can create one directly; free users need to complete API activation first at token.ppthub.shop/settings/apikeys
• Node.js 18 or later
• A terminal

Step 1 — Install OpenCode

npm install -g opencode-ai

China users can use a mirror registry for faster downloads:

npm install -g opencode-ai --registry=https://registry.npmmirror.com

Verify the installation:

opencode --version

Step 2 — Configure Token101 as the Provider

OpenCode is configured through a opencode.json file in your project directory or home directory.

Option A — Project-level config (recommended)

Create opencode.json in your project root:

{
  "provider": {
    "openai": {
      "name": "Token101",
      "apiKey": "YOUR_TOKEN101_API_KEY",
      "baseURL": "https://token.ppthub.shop/api/v1"
    }
  },
  "model": "openai/gpt-5.2"
}

Replace YOUR_TOKEN101_API_KEY with your actual key.

Option B — Environment variables

export OPENAI_API_KEY="your-token101-api-key"
export OPENAI_API_BASE="https://token.ppthub.shop/api/v1"

To persist these, add them to your shell profile (~/.bashrc, ~/.zshrc, etc.).

Step 3 — Verify the Connection

Run a quick test from your terminal:

opencode "Reply with exactly: TOKEN101_OK"

You should see TOKEN101_OK in the response. If you do, OpenCode is connected to Token101.

Usage

Start an interactive session

opencode

OpenCode will open an interactive chat in your terminal. It automatically reads your project's files and can reference them in conversation.

Ask a one-shot question

opencode "Explain what this function does"

Reference specific files

opencode "Review src/lib/billing.ts for potential bugs"

Generate code

opencode "Add input validation to the createUser function in src/handlers/users.ts"

Available Models

You can use any Token101-supported model by updating the model field in your config:

See Supported Models for the full list.

How It Works

OpenCode sends requests to the OpenAI Chat Completions endpoint (/v1/chat/completions). Token101 receives these requests, validates your API key, routes to the appropriate model provider, and returns the response.

OpenCode → POST /api/v1/chat/completions → Token101 → Model Provider → Response

Token101's Chat Completions API is fully OpenAI-compatible, so OpenCode works without any special configuration beyond pointing it at the Token101 base URL.

Troubleshooting

Error: 401 Unauthorized

Your API key is invalid or not set. Verify:

# If using environment variables
echo $OPENAI_API_KEY

# Should print your key, not empty

Double-check the key in your opencode.json matches what's shown in Settings → API Keys.

Error: 404 Not Found

The base URL is incorrect. Make sure it's exactly:

https://token.ppthub.shop/api/v1

Not https://token.ppthub.shop/api (missing /v1).

Model not found

Make sure the model name in your config matches a model from the Supported Models page. Model names are case-sensitive.

Slow responses

For long code tasks, responses can take 30–60 seconds. This is normal. If you're consistently seeing timeouts, try:

• Using a lighter model like claude-haiku-4-5-20251001 for quick tasks
• Breaking large tasks into smaller requests

Next Steps

• Supported Models — see all available models
• Streaming — learn how streaming works
• Claude Code — alternative CLI tool using Anthropic Messages API
• API Reference — full endpoint documentation