Claude Code costs up to $200 a month. Goose does the same thing for free.

The AI coding revolution has a price tag, and it’s starting to piss people off.

<a href="https://write.allwinchina.org/ai-tools/claude-code/" title="Claude Code review”>Claude Code, Anthropic’s terminal-based AI agent that writes, debugs, and deploys code autonomously, is genuinely impressive. But the pricing—$20 to $200 a month depending on usage—has sparked a quiet rebellion among the very developers it’s supposed to help.

Enter Goose. It’s an open-source AI agent from Block (the company formerly known as Square), and it does almost everything Claude Code does, but runs entirely on your local machine. No subscription. No cloud dependency. No rate limits that reset every five hours. Just code.

“Your data stays with you, period,” said Parth Sareen, a software engineer who demoed the tool during a livestream. That’s the core appeal: complete control over your AI-powered workflow, including the ability to work offline. Yes, even on an airplane.

The project has blown up. Goose now has over 26,100 stars on GitHub, 362 contributors, and 102 releases since launch. The latest version, 1.20.1, shipped on January 19, 2026. That’s a development pace that rivals commercial products.

The Claude Code pricing mess

To understand why Goose matters, you need to understand why developers are so frustrated with Claude Code.

Anthropic’s free plan gives you zero access to Claude Code. The Pro plan, at $17 per month with annual billing (or $20 monthly), limits you to 10 to 40 prompts every five hours. If you’re doing serious work, you’ll burn through that in minutes.

The Max plans—$100 and $200 per month—offer more headroom: 50 to 200 prompts and 200 to 800 prompts respectively, plus access to Claude 4.5 Opus. But even these premium tiers come with restrictions that have the developer community fuming.

In late July, Anthropic introduced weekly rate limits. Pro users get 40 to 80 hours of Sonnet 4 usage per week. Max users at the $200 tier get 240 to 480 hours of Sonnet 4, plus 24 to 40 hours of Opus 4. Nearly five months later, people are still angry.

The problem? Those “hours” aren’t actual hours. They’re token-based limits that vary wildly depending on codebase size, conversation length, and complexity. Independent analysis suggests the actual per-session limits translate to roughly 44,000 tokens for Pro users and 220,000 tokens for the $200 Max plan.

“It’s confusing and vague,” one developer wrote in a widely shared analysis. “When they say ’24-40 hours of Opus 4,’ that doesn’t really tell you anything useful about what you’re actually getting.”

The backlash on Reddit and developer forums has been fierce. Some users report hitting their daily limits within 30 minutes of intensive coding. Others have canceled their subscriptions entirely, calling the new restrictions “a joke” and “unusable for real work.”

Anthropic has defended the changes, claiming the limits affect fewer than five percent of users and target people running Claude Code “continuously in the background, 24/7.” But they haven’t clarified whether that’s five percent of Max subscribers or five percent of all users. That distinction matters a lot.

How Block built a free AI coding agent that works offline

Goose takes a radically different approach.

Built by Block, the payments company led by Jack Dorsey, Goose is what engineers call an “on-machine AI agent.” Unlike Claude Code, which sends your queries to Anthropic’s servers, Goose runs entirely on your local computer using open-source language models you download and control yourself.

The project’s documentation describes it as going “beyond code suggestions” to “install, execute, edit, and test” code autonomously. In practice, that means Goose can clone repos, run tests, fix bugs, and even deploy applications—all without sending your code to a third-party server.

This is a big deal for anyone working with sensitive codebases. Financial institutions, healthcare companies, and defense contractors have been understandably nervous about feeding proprietary code into cloud-based AI systems. Goose eliminates that concern entirely.

The trade-offs

Let’s be honest: Goose isn’t a perfect replacement. Running large language models locally requires serious hardware. You’ll want a GPU with at least 8GB of VRAM, ideally more. The open-source models available today aren’t quite as capable as Claude 4.5 Opus for complex reasoning tasks.

But the gap is closing fast. Models like Llama 3.1 70B and Qwen 2.5 72B are remarkably good for coding tasks, and they run entirely offline. For most everyday development work—writing functions, debugging errors, refactoring code—they’re more than adequate.

And let’s talk about the cost argument. At $200 a month, Claude Code Max costs $2,400 a year. That’s a significant expense for individual developers and small teams. Goose, being free, pays for itself the moment you install it.

What this means for the AI coding market

This situation reminds me of what happened with GitHub Copilot. When Microsoft started charging $10 a month, alternatives like Tabnine and Codeium (now Cursor) gained traction. But those were still cloud-based services with their own limits.

Goose represents something different: a genuinely free, no-strings-attached option that puts control back in the developer’s hands. The fact that it comes from Block, a company with real resources and engineering talent, gives it staying power that many open-source projects lack.

I suspect we’ll see more companies following this model. The AI industry has been obsessed with subscription revenue, but there’s a growing segment of developers who value privacy and control over convenience. Goose proves you can have both.

For now, if you’re frustrated with Claude Code’s pricing and rate limits, Goose is worth a serious look. It’s free, it works offline, and it won’t send your code to someone else’s servers. In an industry that keeps finding new ways to charge you, that’s refreshing.