Anthropic's Claude Code, priced up to $200/month, faces a serious challenge from Goose, Block's free and open-source AI coding agent that runs entirely on your local machine.
Paying $200/month for Claude Code sting? Anthropic's pricing structure, with restrictive rate limits, has developers searching for alternatives. Block (formerly Square) offers one: Goose, a free, open-source AI agent that replicates Claude Code's functionality without the subscription fees or cloud dependency.
Anthropic's Claude Code offers AI coding assistance as part of its paid subscription tiers, ranging from $20 to $200 per month with usage limitations. Many developers found these limits restrictive, especially the ambiguous 'hours' of usage that translate to token-based limits (around 44,000 tokens for Pro users). Block's Goose provides similar AI coding functionality but runs locally on your computer using open-source language models. No subscription fees. No rate limits. According to VentureBeat AI, Goose boasts over 26,100 stars on GitHub, highlighting its popularity.
For SMBs, especially those with limited budgets, Goose offers a powerful, cost-effective alternative to expensive AI coding subscriptions. Developers can leverage AI assistance without worrying about monthly fees or usage caps. The ability to run locally also enhances data privacy and security, a critical consideration for businesses handling sensitive information. Imagine cutting $2400/year from your software budget without sacrificing AI coding capabilities.
Goose provides 'genuine autonomy, model agnosticism, local operation, and zero cost,' directly impacting an SMB's bottom line.
Ready to ditch the expensive AI coding subscriptions? Here's how to get started with Goose:
1. **Install Ollama:** Download and install Ollama from ollama.com. This tool simplifies running large language models locally. 2. **Download a Model:** Use Ollama to download a compatible language model, such as Qwen 2.5, which offers strong tool-calling support: `ollama run qwen2.5`. 3. **Install Goose:** Goose is available as a desktop application and command-line interface. Download it from the GitHub releases page (github.com/block/goose). 4. **Configure the Connection:** In Goose, configure the provider to use Ollama, setting the API Host to http://localhost:11434.
Claude Code's premium pricing forces a choice: pay for top-tier AI coding or find alternatives. Goose offers a compelling third option: free, local, and open-source. While it requires more technical setup, the cost savings and enhanced privacy make it a worthwhile consideration for SMBs seeking AI assistance without breaking the bank. It's time to consider bringing your AI coding in-house.
Goose requires a computer with sufficient RAM to run large language models. Block recommends at least 32GB of RAM for larger models, but smaller models like Qwen 2.5 can run effectively on systems with 16GB of RAM.
Goose is model-agnostic and supports a variety of language models, including Anthropic's Claude models (if you have API access), OpenAI's GPT models, Google's Gemini, and open-source models like Meta's Llama series, Alibaba's Qwen, and Google's Gemma.
Yes, one of the key benefits of Goose is its ability to run entirely offline when using local models. This allows developers to work in areas with limited or no internet connectivity without interruption.
While Goose offers similar core functionality, Claude Code's flagship models (like Claude 4.5 Opus) may offer superior performance for complex coding tasks. Goose is a great option for cost or privacy, but benchmark your workflows directly.
Sources
VentureBeat AIBusinessHacks.ai maintains editorial independence. We do not accept payment for coverage. If we recommend a tool, it's because we believe it delivers real value to SMB operators.