Skip to main content

OpenCode

Integrate OpenCode with AI Studio to run models from the command line or OpenCode Studio.

This quickstart guide shows you how to integrate OpenCode with Hyperstack AI Studio using minimal configuration. You’ll install OpenCode, authenticate with AI Studio, and run a model from the command line or OpenCode Studio. This setup enables you to use custom or fine-tuned models from Hyperstack inside your OpenCode workflows.

Full Guide

For the complete guide that includes setup theory, agent design, and workflow demos, see Integrating Hyperstack LLM API with OpenCode.

Why Integrate OpenCode with Hyperstack AI Studio​

OpenCode is a cross-platform CLI and interactive studio for building AI agents and prompt workflows. It supports OpenAI-compatible providers and allows developers to connect custom or fine-tuned models with minimal setup.

Hyperstack AI Studio acts as a managed backend for model inference. It hosts a catalog of base and fine-tuned models accessible via an OpenAI-compatible API. Together, OpenCode and Hyperstack let you:

  • Use custom LLMs from Hyperstack inside OpenCode with no code changes.
  • Combine agentic workflows in OpenCode with powerful inference from GPU-hosted models.
  • Rapidly test prompts, execute tasks, and build AI applications.

To learn more about how this integration works, see the full guide.

This integration gives you full control of the frontend (OpenCode agents and prompts) and the backend (Hyperstack-hosted models).

How to Set Up and Connect OpenCode to AI Studio​

  1. Install OpenCode​

    Install the CLI tool for building and running LLM-powered workflows. For more installation options and details, see the full guide.

    Option A: via Homebrew (macOS/Linux)

    brew install sst/tap/opencode

    Option B: via npm (Cross-Platform)

    npm install -g opencode-ai

  2. Get API Credentials from Hyperstack AI Studio​

    Retrieve the base URL, model ID, and API key from Hyperstack. For more information about selecting models and generating credentials, see the full guide.

    a. Log in to the Hyperstack Console
    b. Open the AI Studio Playground and select a model (e.g. openai/gpt-oss-120b)
    c. Open the API tab to copy the Base URL and Model ID
    d. Go to API Keys and generate a new key
    e. Copy and securely store the API key

  3. Authenticate OpenCode​

    Link your Hyperstack API key to OpenCode by selecting it as a custom provider. For more information about the login flow, see the full guide.

    opencode auth login
    • Scroll to the bottom and select Other
    • Use hyperstack as the provider ID
    • Paste in your Hyperstack API key

    OpenCode Auth UI

  4. Configure opencode.json​

    Set up your OpenCode config file to define Hyperstack as a provider. For config options and tips, see the full guide.

    {
      "$schema": "https://opencode.ai/config.json",
      "provider": {
        "hyperstack": {
          "npm": "@ai-sdk/openai-compatible",
          "name": "Hyperstack AI",
          "options": {
            "baseURL": "https://console.hyperstack.cloud/ai/api/v1",
            "apiKey": "{env:HYPERSTACK_API_KEY}"
          },
          "models": {
            "meta-llama/Llama-3.1-8B-Instruct": {
              "name": "openai/gpt-oss-120b"
            }
          }
        }
      }
    }

  5. Set Your Environment Variable​

    Export the API key to your terminal session so it can be accessed securely.

    export HYPERSTACK_API_KEY="your_hyperstack_api_key_here"
    echo $HYPERSTACK_API_KEY

  6. Verify Setup​

    Confirm that OpenCode recognizes Hyperstack as a valid provider.

    opencode auth list

    Ensure that Hyperstack is listed.

    OpenCode Auth List

  7. Start OpenCode Studio​

    Launch OpenCode’s interactive interface and test your setup. For more test workflows and troubleshooting steps, see the full guide.

    opencode

    Send a prompt like:

    hi

    You should receive a response from your Hyperstack model.

    OpenCode Studio Interface

  8. Test Code Generation​

    Try generating a simple HTML file to verify LLM output.

    create an HTML file of h1 tag

    Expected Output:

    <!DOCTYPE html>
    <html>
      <head></head>
      <body>
        <h1>Hello, world!</h1>
      </body>
    </html>

  9. Optional: Use VSCode Extension​

    Optionally, install the OpenCode VSCode Extension to streamline workflows with agent support.

    VSCode Extension Install


Back to top

On this page