Anthropic Claude has been having its moment lately. Shortly after being labeled a supply chain risk by the Pentagon, it shot to the top of the Apple App Store for the first time, dethroning OpenAI’s ChatGPT. According to reports, free active users on Claude have increased by over 60% since the beginning of 2026, and the number of people signing up daily has quadrupled.

Given all this momentum, more and more people are discovering that Claude has a lot more to offer than just the basic chatbot experience. One of those features is Claude Code, a terminal-based agentic tool that lets you interact with Claude directly from the terminal. While it’s been one of my favorite AI tools lately, it isn’t for everyone. If you happen to be one of them, there’s, fortunately, a free, completely open-source alternative to Claude Code that just so happens to work with almost everything.

OpenCode is a completely free and open-source Claude Code alternative

Same vibes, zero price tag

Given that Claude Code is a feature within Claude, it’s no surprise that it’s locked to Anthropic’s ecosystem. When you're using Claude Code, you're tied to Anthropic's own models like Claude Opus, Sonnet, and Haiku, and you can't really experiment much beyond what's supported out of the box. For instance, if you prefer using local LLMs instead of cloud-based LLMs, Claude Code isn't for you. Claude Code is also currently limited to paid subscribers only.

You need either a Pro or Max plan, meaning you can't really try it out without committing financially first. Ultimately, it's a tough sell if you're just curious about what agentic coding feels like. OpenCode, though, takes a completely different approach. It's free to use, open-source, and it works with over 75 different AI providers, including GPT, Claude, Gemini, and more.

Similar to Claude Code, it's available in a command-line terminal interface too, which means it can plug right into your codebase and handle the same kind of tasks: writing code, running terminal commands, testing your code, debugging issues, and making changes across multiple files at once.

To use OpenCode, you just need a command-line interface (can be the stock macOS Terminal, PowerShell, Command Prompt, or any terminal emulator you already use) and you're good to go. To install OpenCode, simply run the following command in your terminal:

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

You can also download it through npm, Homebrew on macOS, or Scoop and Chocolatey on Windows. You'll find the command to run on this OpenCode's documentation.

OpenCode works with whatever you're already paying for

One less subscription in your life

Once it's installed and all setup, you simply need to run the /models command to select the model you'd like to use. For some LLM providers, you can authenticate directly with your existing subscription.

For example, if you have a ChatGPT Plus or Pro subscription, you can sign in and use it right away. The same goes for GitHub Copilot subscribers. There's also an option to use your Claude Pro or Max subscription, though it's worth noting that this isn't officially supported by Anthropic.

For everything else, you can plug in an API key from whichever provider you prefer and get started that way. And if you'd rather keep everything on your own machine, OpenCode supports local models through tools like Ollama and LM Studio too!

Using OpenCode feels a lot like Claude Code

Familiar? That's the point

With Claude Code, you essentially give Claude permission to view and make changes to your codebase directly from the terminal. You describe what you want in plain English, and Claude goes off and does it for you. OpenCode works the exact same way. Once you've connected an LLM of your choice, you can ask it to explain your code, add features, fix bugs, and essentially anything you'd typically do in a coding session.

Given I'm not a developer by profession and my coding skills admittedly aren't all that strong, one of my favorite Claude Code features to use when vibe-coding is its Plan mode. That's a read-only mode where Claude can't make any changes to your codebase. Instead, it just thinks through the problem and tells you exactly how it would approach it before writing a single line of code. I love this because when you're using AI to code, it's extremely easy to get lost in the weeds and lose track of what you're actually trying to build.

If you begin with a vague idea and leave it up to the LLM, chances are you'll end up with a result where you'll need to do a lot of the cleanup yourself afterward. Given that Plan mode lets you map everything out beforehand, it makes the entire process feel a lot more controlled and intentional. Fortunately, OpenCode has this too. All you need to do is hit Tab to switch to Plan mode and walk through what you're trying to build. When you're happy with it, hit Tab again to switch to Build mode and let it handle the rest.

Beyond Plan and Build mode, OpenCode has a bunch of other features that make the experience feel polished. For instance, you can use the @ symbol to fuzzy search for files in the project and reference them directly in your prompts. You can also drag and drop images right into the terminal, which is incredible if you have a mockup you'd like OpenCode to reference. You can undo changes by typing /undo, and similarly use /redo to bring them back.

One feature I think is underrated is session sharing — you can type /share, and OpenCode will generate a link to your entire conversation. Finally, OpenCode is super customizable. You can pick a theme, create custom commands, code formatters, and more.

OpenCode is absolutely worth trying

Whether you already use Claude Code or haven’t had the chance to try it, I think OpenCode is absolutely worth trying. You get the same structure as Claude Code, similar capabilities, and the option to use models other than Anthropic. If you're curious about agentic coding but don’t want to commit to a paid plan, OpenCode is the perfect way in. And since it's open-source, it's only going to get better.