Claude Code vs Goose: Free AI Agent — 26,100 GitHub Stars

Block (the company formerly known as Square) just open-sourced Goose — an AI coding agent (a program that writes, debugs, and runs code directly on your computer) that does everything Anthropic's Claude Code does, for exactly $0. Meanwhile, a 30-person startup called Railway just raised $100 million after cutting one customer's cloud bill from $15,000 to $1,000 per month. Both stories point to the same shift: the AI tools that defined 2025 now face free, faster, and leaner alternatives.

If you're paying $200 a month for Claude Code, or managing a cloud deployment pipeline (the automated process that takes your code and puts it live on the internet) that grinds through 2–3 minutes per push, this week handed you two serious reasons to reconsider.

The $0 Tool That Does What $200/Month Claude Code Does

Claude Code, Anthropic's terminal-based AI coding agent, costs between $20 and $200 per month depending on usage intensity. It connects to Anthropic's cloud servers (remote computers run by Anthropic that process your requests), which means your code travels off your machine every time you ask it a question.

Goose, built by Block (the fintech company behind Cash App), handles the same workload — write code, fix bugs, run terminal commands, chain complex development tasks — but runs entirely on your local machine. No cloud account. No subscription. No rate limits.

• Price: $0 (Goose) vs $20–$200/month (Claude Code)
• Rate limits: None (Goose runs locally) vs 5-hour resets (Claude Code, cloud-based)
• Data privacy: Your code never leaves your machine (Goose) vs cloud-hosted processing (Claude Code)
• Community traction: 26,100+ GitHub stars, 362 contributors, 102+ releases
• Internet required: No (Goose) vs Yes (Claude Code)

Parth Sareen, a developer who demonstrated Goose in a public livestream, put it in one line: "Your data stays with you, period." At a time when AI companies face growing scrutiny over what they do with code uploaded to cloud services, that distinction carries real weight for teams working on proprietary software.

The 30-Person Startup That Just Cut Fortune 500 Bills by 87%

While Goose grabbed developer attention, Railway — a cloud deployment platform (a service that takes your code and makes it run on the internet) founded in 2020 — closed a $100 million Series B funding round led by TQ Ventures, with participation from FPV Ventures, Redpoint, and Unusual Ventures.

What makes Railway unusual isn't the fundraise. It's the team: 30 employees generating tens of millions in annual revenue, with zero marketing spend. The platform has 2 million developers, processes over 10 million deployments monthly, and routes more than 1 trillion requests through its edge network (the system of globally distributed servers that accelerates internet traffic). Revenue grew 3.5x last year and continues expanding at 15% month-over-month.

The G2X case study shows what Railway actually delivers. G2X is a platform serving 100,000 federal contractors. After migrating to Railway from traditional cloud infrastructure:

• Monthly infrastructure bill: $15,000 → $1,000 (an 87% reduction)
• Deployment speed: 7x faster than the previous setup
• Engineering time: Week-long infrastructure tasks now take a single day

G2X CTO Daniel Lobaton said: "The work that used to take me a week on our previous infrastructure, I can do in Railway in like a day."

Railway's pricing is unusually transparent for the cloud industry:

Memory:  $0.00000386 per GB-second
Compute: $0.00000772 per vCPU-second
Storage: $0.00000006 per GB-second

For comparison, AWS adds data transfer fees, IAM complexity (the permission management system that typically requires a dedicated engineer to configure and maintain), and a billing dashboard with 40+ line items on top of comparable base rates. Railway eliminates most of that overhead by design.

Why Railway Left Google Cloud to Build Its Own Data Centers

In 2024, Railway made a counterintuitive decision: it abandoned Google Cloud and began building its own data centers. That's expensive, risky, and almost unheard of for a startup at this stage. But Railway CEO Jake Cooper had a specific engineering reason.

Traditional infrastructure tools like Terraform (software that automates the setup and configuration of cloud servers) take 2–3 minutes per deployment. Railway completes the same operation in under 1 second. That gap matters more than ever because AI coding agents and vibe coding workflows now generate working code in seconds — and a 3-minute deployment cycle becomes the critical bottleneck in any AI automation pipeline.

Cooper explained the problem directly: "When godly intelligence is on tap and can solve any problem in three seconds, those amalgamations of systems become bottlenecks." Building its own infrastructure gives Railway the vertical integration (end-to-end control from hardware to software) needed to sustain sub-second deployments without negotiating with third-party cloud providers over latency or pricing.

Rafael Garcia, CTO of Kernel, put a human cost on the old approach: "At my previous company Clever, which sold for $500 million, I had six full-time engineers just managing AWS. Now I have six engineers total, and they all focus on product. Railway is exactly the tool I wish I had in 2012."

Two Tools, One Signal: AI Automation Is Moving On-Premise

Goose and Railway represent the same underlying trend. Developers are choosing tools they control — software that runs on their hardware, doesn't meter usage by the hour, and doesn't require ongoing subscription payments to stay functional. The alternative is paying $200/month for a cloud-hosted AI assistant while simultaneously paying AWS for deployment pipelines that take 3 minutes per push.

Today, 31% of Fortune 500 companies use Railway — though it's worth noting some of these are individual team deployments, not company-wide infrastructure migrations. Goose has crossed 26,100 GitHub stars with 362 contributors across 102+ releases, well past the experimental stage. Cooper's forecast is bold: "The amount of software that's going to come online over the next five years is unfathomable — we're talking a thousand times more software." That projection is speculative, but even a fraction of that growth would make current cloud pricing models and 3-minute deployment cycles unsustainable for most teams.

You can try Goose today at github.com/block/goose — no account or API key required to get started with basic tasks. Railway's free tier is at railway.app and connects directly to your Git repositories with no Terraform configuration required. For a deeper look at setting up AI-powered development workflows, our automation guides walk through the full process.

Related Content — Get Started | Guides | More News