Тhe revolution in artificial intelligence coding is not without its drawbacks – it can be quite expensive. Claude Code, Anthropic’s AI agent for coding, has garnered attention from developers worldwide, but its pricing, which ranges from $20 to $200 per month based on usage, has caused dissatisfaction among the very users it aims to serve. This has led to the rise in popularity of a free alternative known as Goose, developed by Block, the financial technology company formerly known as Square. Goose offers similar functionality to Claude Code but operates entirely on the user’s local machine, without any subscription fees, cloud dependency, or rate limits. This has attracted developers who value complete control over their AI-powered workflow, including the ability to work offline, even when on an airplane.
Goose has gained significant traction, with over 26,100 stars on GitHub, 362 contributors, and 102 releases since its launch. The latest version, 1.20.1, was released on January 19, 2026, showcasing a development pace that rivals commercial products. For developers frustrated with Claude Code’s pricing structure and usage caps, Goose provides a genuinely free option for serious work, which is increasingly rare in the AI industry.
On the other hand, Anthropic’s new rate limits for Claude Code have sparked a developer revolt. The pricing controversy surrounding Claude Code has led to dissatisfaction among users, particularly with the new weekly rate limits that were announced in late July. These limits, which are token-based and vary depending on the codebase size, conversation length, and code complexity, have been criticized for being confusing and vague. The backlash on Reddit and developer forums has been intense, with some users reporting hitting their daily limits within 30 minutes of intensive coding. Anthropic has defended the changes, stating that they target a small percentage of users running Claude Code continuously in the background, but the lack of clarity has fueled the discontent among developers.
In contrast, Goose takes a different approach to the problem by running entirely on the user’s local machine. Developed by Block, Goose is an on-machine AI agent that can perform complex development tasks autonomously. It offers flexibility by allowing users to connect to various language models, either locally or through APIs. This setup provides users with complete control over their workflow, without the need for subscription fees or external dependencies.
To set up Goose with a local model, users need to install Ollama, a tool for running open-source models locally, and configure the connection with Goose. Running large language models locally requires substantial computational resources, particularly memory, but smaller models with fewer parameters can run effectively on more modest systems. The practical recommendation is to start with a smaller model to test the workflow and scale up as needed.
Goose’s unique value proposition lies in its autonomy, model agnosticism, local operation, and zero cost. While it may not compete with commercial offerings in terms of model quality or speed, it offers developers a free, privacy-preserving alternative with comparable core functionality. The rise of open-source models like Kimi K2 and GLM 4.5 signals a shift in the AI coding tools market, where the quality advantage of premium pricing may diminish over time. Developers now have a choice between premium commercial products like Claude Code and free alternatives like Goose, reflecting the growing demand for tools that respect their autonomy.
In conclusion, Goose represents a significant development in the AI coding tools market, offering developers a free, local alternative to commercial offerings. While it may have its limitations, the tool provides users with a level of control and autonomy that is increasingly rare in the AI landscape. Goose and Ollama are both free and open-source projects, available for download on GitHub and Ollama’s respective websites.