Claude's free plan quietly got way better — here's what actually changed

A few months ago, the free version of Claude felt like a demo. You'd get a handful of messages, an older model, and a polite suggestion to upgrade every time you tried to do anything useful. That's not the case anymore.

Anthropic has quietly turned Claude's $0 tier into something people are actually building workflows around. No trial countdown, no credit card prompt, no watered-down model hiding behind a paywall. Just a real version of Claude that happens to cost nothing — with some very real limits attached.

So is this actually a good deal, or is it a free sample designed to get you hooked before you hit a wall? Let's break down exactly what you get, what you don't, and whether you still need to pull out your credit card.

What does Claude's free plan actually include now?

The headline change is the model itself. Free users aren't stuck with a stripped-down version anymore — the free tier runs Sonnet 4.5, with chat, artifacts, web search, and file uploads included, and more recent reporting shows claude.ai has a permanent free plan that runs Claude Sonnet 5, one of the strongest models Anthropic ships.

That's a genuinely big deal. A year ago, this level of model quality was locked behind a $20/month subscription. Now it's the default for anyone who signs up with just an email address.

Beyond the model upgrade, here's what's bundled into the free tier based on Anthropic's own pricing breakdown:

Web search built directly into chat, so Claude can pull current information instead of working off a training cutoff
File uploads and image analysis, letting you drop in PDFs, screenshots, or documents for Claude to read
Artifacts, the interactive canvas for code, documents, and visual outputs
Memory across conversations, so Claude can reference things you've discussed before
Code execution, for running Python in a sandboxed environment
Desktop extensions and app connectors, plus Skills
Up to five Projects, Anthropic's persistent workspace feature for organizing related chats and files

You can check the current breakdown yourself on Claude's official pricing page, since Anthropic updates this table fairly often as new features roll out.

How many messages do you actually get for free?

This is where things get murky, and most "guides" online are guessing. Anthropic doesn't publish a fixed message count for the free plan — what it does not have is a fixed message count or a published context number. Instead, usage is metered dynamically based on conversation length, server demand, and how much each message actually costs to process.

That said, independent testing across multiple sources lands on a rough range. Free users typically get around 15-40 messages per 5-hour window with Claude Sonnet 5 before hitting limits, depending on conversation length and current server load.

A few things eat through that budget faster than normal chatting:

Long documents or file uploads cost more tokens than a quick question, since the limit is token-based, not message-based — a prompt that includes a file upload or a long document costs significantly more than a short question.
Web search and web fetch count against the same pool. Asking Claude to pull and summarize several articles in one session will burn through your limit much quicker than a normal back-and-forth conversation.
Peak hours can shrink your effective quota further, since stronger models consume more than smaller models, and peak hours may get a smaller effective quota while higher plans get priority.

When you hit the ceiling, Claude shows a limit notice and you wait for the rolling window to reset. Unlike paid plans, free has no option to buy your way past the wall with usage credits — you simply wait it out.

What's still locked behind a paywall?

The free plan is generous, but Anthropic has drawn a clear line around a handful of features that stay Pro-only. As of the most recent pricing table, free covers chat, web search, memory, files and code execution, connectors, Artifacts, Skills and up to five Projects — but it's missing a growing list of heavier tools.

Here's what requires a paid plan:

Claude Code, Anthropic's terminal-based coding agent. Claude Code requires a Pro, Max, Team, Enterprise, or Console account — Anthropic's setup docs state it plainly: the free Claude plan does not include Claude Code access.
Claude Opus, the most capable (and most expensive) model in the lineup, reserved for Pro and above.
Research mode, which runs structured, multi-source investigations rather than a single search. This lets Claude conduct structured, multi-source investigations — not just a single web search, but a full synthesis of multiple sources into a coherent report — and it's a Pro-exclusive tool.
Claude Design, Slides, and Docs, newer productivity tools that launched in beta on paid plans only.
Claude in Chrome and Claude for Microsoft 365/Outlook, the browser and office integrations.
Unlimited Projects — free caps you at five, while Pro removes that ceiling entirely.
Priority access during high-traffic periods, which matters more than people expect once you're relying on Claude during busy hours.

If your work leans toward actual software development with an autonomous coding agent, or you need Opus-level reasoning for complex tasks, that's where the free plan stops being enough.

Is Claude Pro still worth $20 a month?

It depends entirely on how you use Claude, and the honest answer is: for a lot of casual users, no — not anymore.

If you're using Claude for everyday writing help, brainstorming, quick research questions, or a second opinion on something you're already working on elsewhere, free Sonnet covers that ground comfortably. You get Claude Free with Sonnet access, 30-100 messages per day, Projects, Artifacts, app connectors, web search, and memory features — which is more than most people need for daily use.

Pro starts making sense once you hit a specific wall rather than a general feeling that "it would be nice to have more." The clearest signals you've outgrown free:

You're coding regularly and want Claude Code running in your terminal rather than copy-pasting snippets into a chat window.
You're hitting the message limit multiple times a week, not just occasionally during a busy afternoon.
You need Opus-level reasoning for genuinely hard problems — complex debugging, nuanced analysis, or long multi-step reasoning chains.
You run more than five ongoing Projects and need that workspace structure to stay organized.
You want Research mode for deep, multi-source investigations instead of single web searches.

Pro bumps you up significantly on usage too — per-session usage vs Pro is 1x, with free getting at least 5x less, meaning Pro users get roughly five times the headroom per five-hour window compared to free accounts.

What if you already pay for another AI chatbot?

This is the scenario a lot of people are actually in right now. If you're already paying for ChatGPT Plus or another subscription for your heavy daily work, there's a real argument for keeping Claude on the free tier as your second opinion tool.

Free Claude with Sonnet is strong enough to catch mistakes in a draft, debug a tricky piece of code, or sanity-check an idea — without adding a second $20/month charge to your stack. You keep one paid subscription for your primary workflow and use Claude's free tier for everything else. That's arguably the smartest way to use the current landscape: spread your usage across tools instead of assuming you need every premium tier available.

How do you get started with Claude's free plan?

Setting up is straightforward. Head to claude.ai and sign up with an email address — no credit card required at any point. Once you're in:

Start a new chat and you'll automatically be on Sonnet, the current free-tier default model.
Toggle on web search when you need current information rather than relying on training data alone.
Create a Project if you're working on something ongoing — you get up to five before hitting the cap.
Upload files directly into chat for document analysis, image review, or code review.
Keep an eye on the limit notice. If you're hitting it constantly, that's your signal to evaluate Claude Pro rather than fighting the rolling window.

For the full, current breakdown of what's included on each tier, Anthropic's pricing page is the most reliable source — third-party comparison sites update slower and sometimes quote outdated limits.

Is free Claude enough for you?

The real test isn't whether Claude's free plan sounds impressive on paper — it's whether your actual usage pattern fits inside a rolling five-hour window with a capped model lineup. Light to moderate users will barely notice the limits. Developers, researchers, and anyone running Claude as their primary daily tool will hit the ceiling fast and know exactly when it's time to pay.

Try running your normal week on the free tier before assuming you need to upgrade. You might find the $20 you've been spending was solving a problem you don't actually have anymore.