Is Claude Code Free? Free Tier, Credits, and Budget Options Explained

Anthropic's Claude for Open Source program gives qualifying maintainers six months of Claude Max 20x access, valued at $1,200, at zero cost. That is one of at least five paths to Claude Code that cost less than the $20 per month Pro subscription. Yet most articles answering "is Claude Code free?" stop at a single word: no.

The accurate answer takes more than one word. Claude Code is not included on the free Claude.ai tier. You cannot open a free Claude account, install the CLI, and start coding. The Anthropic quickstart documentation lists three qualifying account types: a Claude subscription (Pro, Max, Team, or Enterprise), an Anthropic Console account with API credits, or access through a supported cloud provider like AWS Bedrock or Google Vertex AI.

But "not free" and "always $20 per month" are not the same thing. Depending on your situation, you may be able to use Claude Code without spending anything, or spend less than a subscription by paying per token through the API. The rest of this guide breaks down every path, ranked from cheapest to most expensive, so you can pick the one that fits your budget and usage pattern.

The free Claude.ai tier gives you a capable chatbot. You can ask coding questions, paste snippets, generate functions, and download artifacts. But you are always the middleman: you copy the code, paste it into your editor, run the tests, and report back.

Claude Code removes that loop. It is a terminal-native AI agent that reads your project files, edits code in place, runs shell commands, manages git operations, and connects to external tools through the Model Context Protocol (MCP). When you tell Claude Code to "fix the failing test in auth.ts," it finds the file, reads the error, writes the fix, and runs the test suite to confirm. You do not touch your editor.

The feature gap between free chat and Claude Code is not incremental. Free chat is a conversation. Claude Code is a coding partner with direct access to your development environment.

The Pro subscription also unlocks Claude Cowork, a companion feature that lets Claude work alongside you on longer tasks directly from your desktop. Both Claude Code and Claude Cowork are completely absent from the free tier.

That distinction matters because it shapes how you evaluate the price. You are not paying for a slightly better chatbot. You are paying for a different category of tool.

Here is every way to get Claude Code access, ranked from cheapest to most expensive.

Anthropic Console Free Credits ($0)

When you create a new account at console.anthropic.com, Anthropic provides roughly $5 in free API credits. No credit card is required. These credits work with Claude Code immediately: install the CLI, log in with your Console account, and start coding.

Five dollars covers an estimated 150 to 300 short Claude Code interactions, depending on the model you select and prompt complexity. That is enough to evaluate whether Claude Code fits your workflow before committing to a subscription. The credits expire, so check your Console dashboard for the exact deadline after creating your account.

Claude for Open Source Program ($0 for 6 months)

Anthropic launched this program in February 2026. Qualifying open source maintainers receive six months of Claude Max 20x, the $200 per month tier, completely free.

Eligibility requirements:

• Primary maintainer of a public repository with 5,000 or more GitHub stars, or 1 million or more monthly npm downloads
• Recent project activity (commits or pull requests) within the last three months
• An exception clause exists for critical infrastructure projects with lower visibility metrics

Up to 10,000 spots are available on a rolling basis, with applications open through June 30, 2026. One important limitation: the program does not include API access. Your ANTHROPIC_API_KEY still incurs separate charges if you use the Console billing path.

Third-Party Model Providers ($0 with caveats)

Claude Code is a client that talks to an API endpoint. You can redirect it to any provider that implements the Anthropic Messages API format. The open source free-claude-code project sets up a local proxy server that routes Claude Code requests to free-tier providers like Google AI Studio (Gemini), Groq, Cerebras, and local models through Ollama or LM Studio.

You can also configure this manually with three environment variables:

export ANTHROPIC_API_KEY=your-provider-key
export ANTHROPIC_BASE_URL=https://provider-api-endpoint/
export ANTHROPIC_MODEL=provider-model-name

The tradeoff is significant: you are running a different model, not Claude. Response quality, tool use reliability, and code generation accuracy will vary by provider. Some free tiers also have rate limits and data usage terms worth reading. But for budget-constrained developers who want the Claude Code interface with a free backend, this path works.

Pro Subscription ($20 per month)

The Pro plan at $20 per month ($17 per month billed annually) is the standard entry point. It includes Claude Code, Claude Cowork, Claude Design, and access to both Sonnet and Opus models. Usage operates on a rolling five-hour window with approximately 44,000 tokens per window, enough for roughly 10 to 40 prompts depending on codebase size and prompt complexity.

As of May 2026, Anthropic doubled the rate limits for Pro and removed peak-hour throttling, making this tier more usable for daily development than it was at launch.

API Pay-As-You-Go (Variable)

If you exhaust the free credits and do not want a subscription, you can add funds to your Anthropic Console account and pay per token. Current API rates per million tokens:

• Haiku 4.5: $1 input, $5 output
• Sonnet 4.6: $3 input, $15 output
• Opus 4.6: $5 input, $25 output

Prompt caching reduces the cost of repeated context to one-tenth of the base input price. The Batch API offers a 50% discount for tasks that do not need real-time responses. For developers who use Claude Code sporadically, a few times per week, pay-as-you-go can cost less than a monthly subscription.

Max Plans ($100 to $200 per month)

Max 5x costs $100 per month and provides roughly 88,000 tokens per five-hour window. Max 20x costs $200 per month with approximately 220,000 tokens per window. Both tiers include priority access during high-traffic periods.

One developer tracked 10 billion tokens over eight months on the Max plan and spent approximately $800, compared to over $15,000 at standard API rates. If you write code with Claude Code for several hours each day, Max pays for itself quickly.

Cloud Providers (Enterprise pricing)

AWS Bedrock, Google Vertex AI, and Microsoft Foundry all support Claude Code through their own billing and authentication. This path makes sense for organizations already committed to a cloud provider and needing consolidated billing, compliance controls, or region-specific data residency.

Store what Claude Code builds

Fast.io gives developers workspace storage with built-in AI search. Upload Claude Code outputs, share with teammates, and query files through the MCP server. Starts with a 14-day free trial.

Start 14-Day Trial

The cheapest option is not always the best one. The right path depends on how often you use Claude Code and whether you need the actual Claude models or just the CLI interface.

Evaluating Claude Code for the first time: Start with the free Console credits. Five dollars is enough to run through a real project task and decide whether Claude Code changes how you work. No subscription commitment, no credit card.

Using Claude Code a few times per week: API pay-as-you-go with Sonnet 4.6 is likely cheaper than Pro. A typical Claude Code session consumes 20,000 to 100,000 tokens. At Sonnet rates, that costs $0.06 to $1.50 per session. If you average fewer than 15 sessions per month, pay-as-you-go beats the $20 subscription.

Daily development: Pro at $20 per month is the right floor. The rolling five-hour window resets continuously, so you can spread heavy sessions across the day without running out. If you consistently hit the rate limit, upgrade to Max 5x before switching to API billing, because subscriptions include a large discount over raw API rates.

Heavy production use or CI/CD pipelines: Max 20x or API with prompt caching. At this volume, the savings that Max provides over raw API rates compound fast. For automated pipelines that do not need interactive responses, the Batch API at half price is worth configuring.

Open source maintainers: Apply for the Claude for Open Source program before paying anything. If your project qualifies, you get the highest tier free for six months.

Whichever path you choose, a few habits reduce your token spend without reducing the quality of Claude Code's output.

Write a CLAUDE.md file. Claude Code reads this file at the start of every session to understand your project structure, coding conventions, and preferred tools. A well-written CLAUDE.md prevents Claude from spending tokens exploring your codebase from scratch each time. The upfront investment of writing one saves tokens on every subsequent session.

Choose models by task complexity. Haiku 4.5 at $1/$5 per million tokens handles simple refactors, renaming, and boilerplate generation well. Sonnet 4.6 covers most development work. Reserve Opus 4.6 for complex architectural decisions and multi-file refactors where you need the strongest reasoning.

Use prompt caching aggressively. When Claude Code re-reads the same files across turns in a conversation, prompt caching reduces those repeated context tokens to one-tenth of the base cost. Longer sessions with consistent context are cheaper per token than many short, disconnected sessions.

Persist outputs in a shared workspace. Claude Code generates files locally, but those files can disappear when you switch branches, reformat a drive, or hand off to a teammate. Uploading outputs to a persistent workspace like Fast.io means the work survives beyond your local machine. Fast.io offers built-in AI search across uploaded files and an MCP server that other agents can query, and every org starts with a 14-day free trial.

Batch non-urgent work. The Anthropic Batch API processes requests at half the standard API rates. If you have a queue of code review tasks, test generation, or documentation updates that do not need real-time responses, routing them through the batch endpoint cuts your cost in half.