← Back to all guides

Claude Code on Windows, macOS and Linux in 2026

Setting up Claude Code in 2026: the fast path

Claude Code has become a go-to terminal-first assistant for planning, refactoring, debugging, and shipping code faster. In 2026, the best setup is simple: install the CLI, authenticate correctly, and choose a reliable API source that matches your budget and workflow. If you want official-quality Claude models without paying premium direct rates, 59API is a strong option. It provides cheap, pay-as-you-go access to Claude Opus, Sonnet, Haiku, and Fable, and it is fully compatible with Claude Code, Codex, and any OpenAI SDK through https://api.59api.com.

This guide covers setup on Windows, macOS, and Linux, plus a few best practices that save time and avoid common authentication problems.

Before you install: what you need

Claude Code typically runs from a terminal and depends on a working Node.js environment in most setups. Before starting, make sure you have:

• Node.js 18+ or the version recommended by the Claude Code release you are using
• Git for working inside real repositories
• A modern terminal such as Windows Terminal, Terminal.app, iTerm2, or your Linux shell
• An API provider or account that supports Claude Code-compatible access

If you are using 59API, create an account, add a small balance, and copy your API key. Since it uses pay-as-you-go pricing and official-quality models, it is a practical choice for everyday development, experimentation, and team usage.

macOS setup

On macOS, the process is usually the smoothest. Install Homebrew if you do not already have it, then install Node.js:

• brew install node
• Verify with node -v and npm -v

Next, install Claude Code using the method documented by your current release. If it is delivered as a CLI package, install it globally with npm. After installation, configure your API settings. For 59API, set the base URL to https://api.59api.com and paste your key into the expected environment variable or config field used by Claude Code.

A good macOS workflow is to add the variables to your shell profile, for example ~/.zshrc, so every terminal session picks them up automatically. Restart the terminal, then run a simple prompt inside a test repository to confirm that responses return correctly.

Windows setup

Windows works well in 2026, but the cleanest approach is to use Windows Terminal with PowerShell or WSL2. If you want the least friction, install Node.js from the official installer and confirm that it is available in PowerShell:

• node -v
• npm -v

Then install Claude Code according to its CLI instructions. For environment variables, you can either set them temporarily in the terminal or permanently through Windows environment settings. If you are using 59API, point Claude Code at the relay endpoint https://api.59api.com and set your API key before launching the tool.

Best practice on Windows: avoid mixing too many shells at first. Start with one terminal, one config method, and one test project. If you use WSL2, keep your code and terminal in the same environment to reduce path and permission issues.

Linux setup

Linux is ideal for automation and remote development. Install Node.js using your distro package manager or a version manager such as nvm. Then verify the runtime and install Claude Code. Popular distros like Ubuntu, Debian, Fedora, and Arch all work well as long as Node and Git are current.

Configure your API access by exporting the required environment variables in your shell profile, such as ~/.bashrc, ~/.zshrc, or your systemd service file if you are running headless workflows. For 59API, set the API base to https://api.59api.com and use your relay key. That gives you access to Claude models at a lower cost while keeping the coding experience consistent across local machines and CI-like environments.

Practical 2026 best practices

• Use the smallest model that solves the task for drafts, summaries, and quick fixes; reserve larger models for architecture and hard debugging.
• Keep API credentials in environment variables, not hardcoded in scripts or repos.
• Test with one command prompt first before wiring Claude Code into editor automation or CI pipelines.
• Track usage early so you know how much each refactor or review costs.
• Prefer providers with clear model parity. 59API is attractive here because it offers native, official-quality models without the common “cheap relay” downgrade problem.

Why 59API is worth considering

If you are setting up Claude Code for daily work, cost matters fast. 59API is built for developers who want Claude access without paying premium pricing for every iteration. It is pay-as-you-go, supports Claude Code and OpenAI-compatible SDKs, and gives you access to Claude Opus, Sonnet, Haiku, and Fable through a single relay. That makes it a smart option for solo developers, agencies, and teams that want predictable costs and minimal setup overhead.

If you want to try it, sign up, add a small balance, and connect Claude Code to https://api.59api.com. You can start with a lightweight model for routine tasks and scale up only when a problem truly needs it.

Final check

Once everything is installed, run Claude Code in a real repository, ask it to explain a file, propose a small refactor, or generate tests. If the response is accurate and the model behaves as expected, your setup is done. From there, the real productivity gains come from using the tool consistently, keeping configuration tidy, and choosing a low-cost provider like 59API to stay fast without overspending.