---
title: "How to Get Claude Code at the Lowest Cost: Plans and Free Routes"
description: "Claude Code installs for free but needs a paid Claude plan or API credits to run. Every plan's price, how the usage limits work, and the few free routes that are real."
author: "Jahanzaib Ahmed"
date: 2026-11-14
category: "implementation"
readingTime: "10 min read"
tags: ["claude code", "anthropic", "claude", "ai pricing"]
canonical: https://www.jahanzaib.ai/blog/is-claude-code-free
source: https://www.jahanzaib.ai
---
# How to Get Claude Code at the Lowest Cost: Plans and Free Routes

Is Claude Code free? No. It installs for free, but it doesn't run on Claude's Free plan, so you need a paid plan (Pro is $20 a month, or $17 a month billed yearly) or pay as you go API credits, per Anthropic's [pricing page](https://claude.com/pricing) on November 14, 2026. In about 10 minutes you'll pick the cheapest plan that fits, install it, and make sure it bills your plan and not an API key.

## Pricing for Claude Code on every plan

These are all the ways to pay, from Anthropic's [pricing page](https://claude.com/pricing) as of November 14, 2026. None of them charges extra for Claude Code.

| Plan | Price | Claude Code | Usage and extras |
| --- | --- | --- | --- |
| Free | $0 | Not included | Limited chat usage only |
| Pro | $20 a month, or $17 a month billed yearly ($200 up front) | Included | More usage than Free, plus optional usage credits |
| Max | $100 a month | Included | 5x Pro usage, plus $100 a month in API credits |
| Ultra | $200 a month | Included | 10x Pro weekly usage, plus $200 a month in API credits; no 5 hour limit on verified accounts |
| Team, Standard seat | $25 per seat a month, or $20 billed yearly | Included | More usage than Pro; 2 to 150 people |
| Team, Premium seat | $125 per seat a month, or $100 billed yearly | Included | 5x a Standard seat |
| Enterprise | $20 per seat a month billed yearly, 20 seat minimum | Included | Usage charged at API rates |
| Console (API) | Pay per token | Included | Billed at standard API prices |

Team plans also pool up to $500 a month in API credits across the team, per the same page. Prices exclude tax and can change, so check the page on the day you buy.

## Before you start

-   A Claude account, the same one you use for chat.
-   A terminal on macOS, Linux or Windows (Windows 10 version 1809 or later), or the Claude desktop app.
-   A project folder you're happy to let an agent read and edit.
-   A rough idea of how many hours a week you'll code with it, since that decides the plan.

## Step 1: Check whether Claude Code is free for you

Before paying, check the three routes that cost nothing. There aren't more, whatever the videos say.

-   **Open source maintainers:** Anthropic's [Claude for Open Source](https://claude.com/contact-sales/claude-for-oss) program gives six months of Claude Ultra at no cost to maintainers of widely used packages, core and active contributors, community builders and maintainers of critical infrastructure. If you don't clearly fit, you can still apply and explain how others depend on your work. When the six months end, the account goes back to its earlier plan, or to Free.
-   **Students and staff:** the pricing page has no individual student discount. Anthropic sells discounted education plans to whole institutions, so ask whether your university has one.
-   **Your employer:** a Team or Enterprise seat already includes Claude Code.

If none of those fit, you're paying. That's fine, because the cheapest plan costs less than most people expect.

## Step 2: Pick the cheapest plan that fits

My rule is simple. Start on the smallest plan and upgrade only when the limits start costing you real time.

-   Trying it out, or a few sessions a week: Pro. It's the cheapest way in, and you can cancel after a month.
-   Daily coding on real projects: Max. If you hit Pro's limits more than once or twice a week, the upgrade costs less than the hours you lose waiting.
-   Agents running for hours, or several sessions at once: Ultra, mainly for the higher weekly limit and no 5 hour window.
-   A team of developers: Team, so seats, billing and admin controls live in one place.
-   Claude Code inside your own scripts or CI: a Console account with API billing and a spend cap.

![Decision tree mapping light use to Pro, daily coding to Max, long agent runs to Ultra and scripts to an API key](https://cdn.sanity.io/images/qajb7q5q/production/d71e4db38dc45eed13c771e5e101942d5bbff411-1376x768.png?w=1200&q=75&auto=format&fit=max)

_Match the plan to how often you code: start on Pro and move up only when the limits cost you time._

I'd pay monthly for the first month. Yearly Pro saves $40 over twelve months ($200 against $240), but only once you know you'll keep using it.

## Step 3: Install Claude Code

On macOS or Linux, open a terminal and run the install command from the [product page](https://claude.com/product/claude-code):

```
curl -fsSL https://claude.ai/install.sh | bash
```

On Windows, open PowerShell and run the command from Anthropic's [Claude Code quickstart](https://code.claude.com/docs/en/quickstart):

```
irm https://claude.ai/install.ps1 | iex
```

The quickstart also gives a CMD command and a WinGet option, `winget install Anthropic.ClaudeCode`, and you don't need to run as Administrator. It recommends Git for Windows so Claude Code can use Bash; without it, Claude Code runs commands through PowerShell. Open a new terminal and run `claude --version` to confirm it installed.

Prefer not to use a terminal? Claude Code also runs in the desktop app, VS Code, JetBrains IDEs and on the web at claude.ai/code, and every one of them signs in with the same account.

## Step 4: Sign in with your plan, not an API key

Run `claude` in your project folder and sign in with your Claude account. Then run `/status`. It shows which login the session is using, and it's the check most people skip.

Anthropic's [support article on using Claude Code with your plan](https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) says that if the `ANTHROPIC_API_KEY` environment variable is set, Claude Code "will use this API key for authentication instead of your Claude subscription." Your plan's usage sits there unused while every token bills the API.

It happens easily. You set the key months ago for a script, forgot it, and now every session bills that key. If `/status` shows a key you didn't mean to use:

1.  **Remove the key.** Delete the `ANTHROPIC_API_KEY` line from your shell profile or `.env` file, then open a new terminal.
2.  **Log in again.** Run `/logout`, then `/login`, and sign in with your Claude account.
3.  **Check once more.** Run `/status` and confirm it now shows your plan.

If you do want API billing for scripts or CI, give that key its own spend limit first. I walk through it in [how to get a Claude API key and cap what it can spend](https://www.jahanzaib.ai/blog/claude-api-key-setup-first-request).

## Step 5: Keep an eye on your usage limits

Paid plans aren't unlimited. The same support article describes two limits: a 5 hour session limit on Pro and Max, which verified Ultra accounts don't have, and a weekly limit on every plan. Max and Ultra add a third: a separate weekly limit just for Fable models.

That Fable limit can run out before the rest of your plan does. If you lean on Fable for long agent runs, you can hit it midweek while normal usage is still left. On Max and Ultra you can turn on usage credits to keep going once it's reached. Anthropic doesn't say whether Fable work also counts toward the general weekly limit, so check `/status` rather than guess.

Claude chat and Claude Code also draw from the same allowance. A long research chat in the morning leaves less room for code in the afternoon. So run `/status` before a big task, not after.

![Diagram showing Claude chat and Claude Code drawing from one shared allowance capped by a 5 hour and a weekly limit](https://cdn.sanity.io/images/qajb7q5q/production/e794cf6b7c3e90ceca2230a2921dccf716f24e3c-1376x768.png?w=1200&q=75&auto=format&fit=max)

_Chat and Claude Code spend the same allowance, so a long morning chat shrinks your afternoon coding time._

When you hit a limit you have three choices. Wait for the reset, which costs nothing. Turn on usage credits, which Pro, Max and Ultra can buy to keep going. Or log in with a Console account and pay API prices. Anthropic doesn't publish exact limit sizes or reset times, so `/status` is the only reliable guide.

## Troubleshooting Claude Code billing and sign in

### Claude Code signs in with the wrong account

Run `/logout`, then `claude update`, restart the terminal and run `claude` again, as the support article suggests. Finish with `/status` to confirm the right plan.

### You see "Usage limit reached · wrapping up"

You hit the 5 hour limit in the middle of a task. Claude Code may finish the current step and then stop. Anthropic says this small allowance counts toward your weekly limit and is still rolling out, so don't plan around it.

### Your Max or Ultra API credits aren't covering Claude Code

**Warning:** The monthly API credits that come with Max and Ultra can't be used in Claude Code or the Claude apps, according to Anthropic's support article. They're for API projects only, so don't count on them for a coding session.

## Free Claude Code proxies and Ollama: what you really get

Most "Claude Code is now free" videos use the same trick. A community proxy sits between the Claude Code app and the model, then sends your requests somewhere else: a local model through Ollama, a free tier on OpenRouter, or NVIDIA NIM. The app is still Claude Code. The model answering you is not Claude.

Three problems come with it:

-   **You lose what you came for.** Claude Code's results come mostly from the Claude model behind it. A small local model in the same app is a very different tool.
-   **It's unsupported.** Anthropic documents [connecting Claude Code to an LLM gateway](https://code.claude.com/docs/en/llm-gateway-connect), but that's for gateways your company runs in front of Claude, not community proxies. Features can break whenever either side updates.
-   **Your code goes through someone else's software.** Your files and prompts pass through a stranger's proxy and, with free cloud tiers, a stranger's servers.

My take: it's fine for learning how a coding agent works or for throwaway side projects. I wouldn't point it at client code or anything with secrets in the repo.

## Was Claude Code removed from the Pro plan?

Briefly, and only for a small group. On April 21, 2026, Anthropic's pricing page stopped listing Claude Code under Pro, [The Register reported](https://www.theregister.com/2026/04/22/anthropic_removes_claude_code_pro/). Anthropic's head of growth called it "a small test on ~2 percent of new prosumer signups" and said existing subscribers weren't affected.

As of November 14, 2026, the pricing page lists Claude Code in every paid plan, Pro included. But Anthropic also said its plans weren't built for how heavily people now use agents, so I'd expect the limits to keep shifting.

## Going further: API billing and when to skip Claude Code

When is the API cheaper? For light, occasional use it can be, because you pay only for what you send. Anthropic's [pricing page](https://claude.com/pricing) lists, per million input and output tokens, $4 and $20 for Opus 5.5 and $2 and $10 for Sonnet 5.5. Haiku 5.5 costs $0.10 and $0.50 on prompts up to 100K tokens, and $0.50 and $2.50 above that. Batch processing takes 50% off.

A rough example shows why plans win for coding. Say one long session on Sonnet 5.5 sends 2 million input tokens and gets 200K output tokens back. That's $4 for input and $2 for output, so $6. Four sessions like that cost more than a month of Pro at $20. [Prompt caching](https://www.jahanzaib.ai/glossary/prompt-caching) cuts repeated context to $0.10 per million cached tokens read, but coding sessions fill a large [context window](https://www.jahanzaib.ai/glossary/context-window) on every turn, so for daily work a flat plan almost always costs less. These numbers are an illustration, and real sessions vary a lot.

If you're choosing a provider for a business agent rather than a coding tool, I compare the two in [OpenAI vs Claude for business AI agents](https://www.jahanzaib.ai/blog/openai-vs-claude-business-ai-agents).

And sometimes you don't need Claude Code at all. If you only want occasional answers about code, chat on the same plan does that without giving an agent access to your files. Before you let it loose on a real repo, run your settings through my [Claude Code permissions checker](https://www.jahanzaib.ai/tools/claude-code-permissions-checker).

Once you're set up, my [Claude Code cheat sheet](https://www.jahanzaib.ai/cheat-sheet/claude-code-cheat-sheet) covers the commands worth knowing, and [how to add MCP servers with claude mcp add](https://www.jahanzaib.ai/blog/claude-mcp-add) is the next step for connecting your own tools.

If you'd rather have an agent built and kept inside a budget for you, that's what I do: [see how I build AI agents](https://www.jahanzaib.ai/agents).

## Frequently asked questions

### Can I use Claude Code for free in VS Code?

No. The VS Code extension signs in with the same Claude account as the terminal app, so it needs a paid plan, a Team or Enterprise seat, or a Console account with API credits. The only ways to run it without paying are Anthropic's open source program, a seat from your employer, or unsupported proxies that swap Claude for another model.

### Does Claude Code cost extra on top of Claude Pro?

No. Claude Code is included in Pro and every other paid plan at no extra charge, and it uses the same usage limits as Claude chat. You only pay more if you turn on usage credits after hitting a limit, or if Claude Code bills an API key because ANTHROPIC\_API\_KEY is set on your machine.

### What happens when I hit my Claude Code usage limit?

Claude Code stops taking new work until the 5 hour or weekly limit resets, and it may finish the current step first. You can wait for the reset at no cost, turn on usage credits to keep going, or log in with a Console account and pay standard API prices. Run /status to see where you stand.

### Can I use my Max plan's API credits in Claude Code?

No. Anthropic's support article says the monthly API credits included with Max and Ultra can't be used in Claude Code or the Claude apps. They're meant for API projects, such as scripts or apps you build yourself. Claude Code on Max draws from the plan's own usage allowance instead, which is five times what Pro gets.

### Is there a Claude Code student discount?

Not for individuals. The pricing page lists no student plan or discount code. Anthropic offers education plans to whole universities at discounted rates, so the practical step is to ask your university's IT team whether it has one. If it doesn't, Pro at $20 a month is the cheapest way in.

### Can I use Claude Code with Ollama or NVIDIA NIM for free?

Yes, through community proxy projects that point the Claude Code app at other models, including local Ollama models and NVIDIA NIM. It costs nothing, but you're no longer using Claude, Anthropic doesn't support it, and features can break when either side updates. I'd keep it for learning and side projects, never for client code.

## Related

- [How to Remove an MCP Server From Claude Code When It Keeps Coming Back](https://www.jahanzaib.ai/blog/claude-remove-mcp-server)
- [Anthropic Scion Benchmarks: 8.8x Faster CI Checkouts, 25x Faster Pulls](https://www.jahanzaib.ai/blog/anthropic-scion-source-control)
- [How to Add MCP Servers to Claude Code With claude mcp add](https://www.jahanzaib.ai/blog/claude-mcp-add)

---

Canonical HTML version: https://www.jahanzaib.ai/blog/is-claude-code-free
