Is Claude Code Free? Free Ways to Use It in 2026

Claude Code is free to install but not free to run. The command-line tool itself is free and open-source, so downloading it costs nothing, but every task it performs calls a Claude model, and that usage is what you pay for. You can cover it with a paid Claude Pro or Max plan, with pay-per-token API credits, or with one of several genuine free workarounds, and which model answers by default now depends on your plan: Claude Sonnet 5 on Pro, Claude Opus 5 on Max and the API.

So the honest answer to “is Claude Code free or paid” is “both, depending on how you run it.” This guide covers whether the free Claude plan includes Claude Code, the four real ways to use it for free, whether it comes with Claude Pro, which model you get, and the best free alternatives. If you just want to try the terminal coding agent without spending anything, one of the paths below will get you there.

The Key Takeaways

• The Claude Code CLI is free and open-source to install; you only pay for the model that runs behind it.
• The free claude.ai plan does not include the Claude Code agent, only the chatbot.
• New API accounts get around $5 in free credits, roughly 150 to 300 short interactions, with no card required.
• You can run Claude Code on free or local models by pointing it at a different endpoint, so the running cost drops to $0.
• Claude Pro ($20/month) and Max both include Claude Code, in the same subscription as the Claude apps. The default model splits by plan: Sonnet 5 on Pro at $2/$10 intro pricing through August 31, 2026, and Opus 5 on Max and the Anthropic API.

Is Claude Code Free or Paid?

Claude Code sits in a confusing middle ground, which is why the question keeps coming up. The CLI itself is free and open-source to install on macOS, Linux, or Windows, and the same is true of the VS Code extension. What is not free is the intelligence behind it, since every command reads files, writes code, and re-checks its own work by calling a Claude model, and those model calls cost money.

The one thing to be clear on is the free plan. The free claude.ai account does not include the Claude Code agent, as Engadget confirmed in its breakdown of what is free in 2026. You can ask coding questions in the free chatbot and copy-paste snippets, but the terminal agent that edits your codebase is gated behind a paid plan or API billing.

That still leaves several ways to run Claude Code without paying full price, which is what most people are really asking about. The next section walks through each one, from free credits to running it entirely on your own hardware.

How to Use Claude Code for Free

There are four real paths to running Claude Code for free or nearly free. Each suits a different type of user, so pick the one that matches how you work.

Free API starter credits

The simplest free path is the Anthropic API. New API accounts get roughly $5 in free starter credits with no credit card required, and Claude Code can run directly on an API key instead of a subscription. That $5 covers about 150 to 300 short interactions, or one to three days of light coding, which is plenty to see whether the tool fits your workflow.

Once the credits run out, you decide whether to top up your API balance or move to a flat plan. For people who only touch Claude Code a couple of days a week, staying on pay-per-token API billing often beats any subscription.

Point Claude Code at a free model

This is the trick the pricing pages leave out. Claude Code does not care which model sits behind the API, so you can set the ANTHROPIC_BASE_URL environment variable to point it at a different endpoint entirely. Aim it at OpenRouter, which offers a rotating set of free models like DeepSeek and Qwen at around 50 requests a day, or at Google AI Studio for a free Gemini API key, and Claude Code works almost identically at $0.

Open-source helpers make this easy. Tools like claude-code-router and the popular free-claude-code project on GitHub handle the translation so you can wire up a free backend in a few minutes. Output quality tracks whichever free model you choose, so it sits below the real Claude, but the price is unbeatable.

Run Claude Code on a local model

If you would rather keep your code off the cloud, point Claude Code at a model running on your own machine. Using Ollama with a coding model like Qwen 3.6 Coder or DeepSeek V4, you get a fully local agentic setup that costs nothing but electricity. Speed and quality depend on your hardware, so this route pays off most if you have a capable GPU sitting idle.

Claude for Open Source

If you maintain a popular open-source project, this is the most generous free deal Anthropic offers. The Claude for Open Source program grants qualifying maintainers six months of Max 20x free, a $1,200 value that includes full Claude Code access. Eligibility leans on hard numbers like 5,000+ GitHub stars or 1 million+ monthly downloads, plus recent commits.

One caveat on timing. The program ran applications through June 30, 2026 on a rolling basis, so it may be closed or paused by the time you read this. Check the application page for current status before counting on it.

Is Claude Code Free With Claude Pro?

Yes. If you already pay for Claude Pro at $20/month, Claude Code is included at no extra cost, and the same is true on the Max plans. Pro bundles the coding agent with the chatbot on a single shared usage budget, so a subscription you may already have for chat also covers the terminal. This covers the common “is Claude Code free for Pro users” and “personal use” questions, since Pro is a personal plan and Claude Code is part of it.

The catch is capacity. Pro gives you a handful of focused coding sessions a day before you hit the rolling usage limit, which is fine for side projects but tight for all-day development. Our guide to Claude’s usage limits explains how that rolling window is measured. Heavy users move up to Max at $100 or $200/month for far more headroom.

Here is the short version of the paid landscape.

This is only the outline. For the full plan-by-plan breakdown, every API rate, and real monthly spend, see our complete Claude Code pricing guide. If you only want the chatbot, our guide to whether Claude AI is free covers the chat side.

Which Model Does Claude Code Run?

There is no single answer any more, because Claude Code picks its default from your account type. On Pro, Team Standard and Enterprise seats the default is Claude Sonnet 5, the same model that powers the free chat plan. On Max, Team Premium, Enterprise pay-as-you-go and the Anthropic API, the default is Claude Opus 5. That split is worth knowing before you compare cost notes with someone on a different plan, because you may not be running the same model at all.

Sonnet 5 is the cheaper of the two. Per Anthropic’s Sonnet 5 announcement, it launched on June 30 and is priced at $2 per million input tokens and $10 per million output through August 31, 2026, after which it rises to $3/$15. That intro rate is why your free API credits stretch further right now than they did on the older models most guides still quote.

If you are on Pro you can still switch up to Claude Opus 5, Anthropic’s top-tier model since July 24, 2026, for the hardest architectural work, though Opus costs more per token and burns a paid budget faster. One practical catch: Opus 5 needs Claude Code v2.1.219 or later, so run claude update if it is missing from your model picker. Lighter tasks can drop to Haiku 4.5 to save even more. Whichever model you pick, the free-versus-paid rules above stay the same.

The Best Free Alternatives to Claude Code

If you mainly want AI help with code and are not tied to Anthropic’s CLI, cheaper all-in-one options exist. Meta shipped a rival terminal agent in August 2026, and our guide to Muse Spark 1.2 and Muse Code covers its pricing and the data trade behind its cheapest tier. Fello gives you Claude alongside ChatGPT, Gemini, Grok, and DeepSeek in a single app, so you can write code, debug, and swap models in one chat without wiring up API keys or environment variables. It includes a Free Compound model that stays free with no message limits for everyday questions, or one flat $9.99/month for every frontier model, which is half the price of Pro and covers five model families instead of one.

For pure coding on a zero budget, the open models are the move. Our roundup of the best free AI for coding tests DeepSeek, Qwen, and Kimi head to head on real programming tasks, and several match paid tools on everyday work. Start there or with the free workarounds above, then upgrade to Claude Code only when you actually hit the ceiling.

Conclusion

Claude Code is free to install but not free to run, and the free claude.ai chat plan does not include the agent. The genuine free paths are the $5 API starter credits, pointing the tool at a free or local model, and the Claude for Open Source program if you qualify and applications are open. If you already pay for Pro at $20/month, Claude Code is bundled in at no extra cost.

Start with the free credits or a free-model setup, see how far they get you, and upgrade only when the limits bite. If you would rather not manage a single-vendor CLI at all, a multi-model app like Fello or a free open coding model may cover you for less. Compare the full field on our best AI models hub.

FAQ