Is Claude free? Yes, with a 5-hour session limit

What the Claude free tier actually gives you

The Claude free tier is a real, usable plan with no card and no trial clock: you sign in and chat, and you stop when you run out of usage for the current window. As of October 2026 the free plan costs $0 and covers chat on the web, on iOS and Android, and in the desktop app, plus web search, memory across conversations, file creation with code execution, Artifacts, Claude Design, Slides and Docs, connections to your apps and tools, and up to five projects. It does not include Claude Code, which starts on Pro. Claude Cowork is on its way to no longer being a separate product: on September 16, 2026 Anthropic announced that Cowork and chat are merging into one Claude. The change is rolling out in stages, Pro, Max and Ultra first, so accounts on the same plan get it at different times, and if your message box still shows separate Chat and Cowork options you do not have it yet. Anthropic says Free plans will follow but has given no date. On Team plans the merged experience is in beta, and on Team and Enterprise it applies only where the Chat and Cowork unified setting is turned on. Anthropic called the merged experience generally available for Enterprise on October 23, 2026, but as of October 27, 2026 its help center says the setting is off by default in existing Enterprise organizations, an Owner or Primary Owner can turn it on early, and it turns on by default on November 16, 2026.

So the honest answer to "how many times can I use Claude for free" is: enough for short daily questions, and not enough for a working session. The free plan is metered by usage inside a rolling five-hour session window and a weekly limit rather than by a fixed lifetime quota, so it refills on its own when each limit resets.

How the free tier is measured, and when it resets

Claude does not sell you a fixed number of messages. It meters how much work your conversation costs, so the count you get is a range and not a number. Two things move it.

• Conversation length. Every turn includes conversation context, so a long chat uses more allowance than a fresh one. Long threads hit the wall faster than several short ones.
• Attachments and tools. A PDF, a spreadsheet, or a web search pulls extra text into the request, and that text is metered like anything else.

When you run out, the interface tells you the reset time in the message itself, and that time is the number that matters. Anthropic also states plainly that limits vary with demand, so a figure copied from a blog post will be wrong on the day you read it. Read your own reset time instead of trusting a table. As of October 31, 2026, Anthropic's pricing page lists both a five-hour session limit and a weekly limit on the free plan. Anthropic also says it may impose other types of usage limits to keep access fair for everyone, and as of October 27, 2026 its pricing FAQ gives weekly and monthly caps or limits on model and feature usage as examples, applied at its discretion. What to do when you hit a Claude usage limit walks through the exact messages and the commands that show your remaining allowance.

One habit saves more free usage than anything else: start a new conversation when you change subject. Carrying an unrelated ten-message history into a new question uses allowance for unrelated context on every turn, for nothing.

What Pro adds that the free plan does not have

The difference is not only a bigger allowance. As of October 2026, Pro is $17 per month billed annually or $20 per month billed monthly, and it opens features the free plan does not carry at all. Pro has a ceiling of its own, and the point where Claude Pro's limits stop you is worth knowing before you assume the upgrade ends the problem.

• Claude Code. This is the agentic coding surface, and it is gated behind Pro. Free also supports desktop extensions and custom MCP connectors, which can give Claude tools for working directly on a codebase, including editing files and Git operations; Free is limited to one custom remote connector. Free is not shut out of code altogether: you can ask Claude to write and fix code in chat, Artifacts are available on every plan including Free and let you create, edit and share code snippets, and file creation with code execution runs code for you. Free can also read a repository: the GitHub integration adds repository files to a chat or project so Claude can answer questions about the code. Claude Code also spends from the same usage window as your chats rather than a separate allowance, and which plans actually include Claude Code covers that overlap along with the point where an API key starts billing you instead.
• Handed-off and scheduled tasks. Cowork used to be a separate surface that Pro unlocked. Since Anthropic's September 16, 2026 announcement it is being folded into ordinary chat, so on accounts that have the merged experience, what Cowork does is available from any conversation. That merge is rolling out in stages, Pro, Max and Ultra first, so two accounts on the same plan can get it at different times; on Team it is in beta where the unified setting is on, and Free is due to follow with no date given. As of October 2026 the Pro feature list on the pricing page names handing off and scheduling tasks, Claude Dashboards (beta), Claude Science, and Claude in Chrome and Microsoft 365. Claude Design, Slides and Docs are no longer Pro additions: Anthropic has taken them out of beta and made them available on every plan, including Free.
• Model choice. The free plan lets you pick between Sonnet and Haiku. Pro adds Opus and opens Fable through usage credits, which matters when you want a cheaper model for routine work and a stronger one for hard problems.
• Unlimited projects and Research. The free plan includes projects too, but caps you at five, and it does not include Research. Pro lifts the project cap and adds Research.
• More usage. The allowance is larger: as of October 2026 the pricing page puts Pro at a minimum of five times the Free usage per five-hour session. It runs on the same rolling session window, so it resets rather than expiring for the month, and Pro also has a weekly limit.

Above Pro, Max costs $100 per month and gives five times the Pro usage every five hours and every week, while Ultra costs $200 per month and gives ten times the Pro usage every week. Both prices are for web subscriptions, mobile pricing may vary, and both plans are billed monthly only. Verified Ultra accounts have no five-hour limit; accounts not yet verified have a temporary limit. Ultra has a weekly limit across all models and a separate weekly Fable limit. Both plans include Fable up to half the weekly allowance, priority access during peak hours, and monthly Claude Platform API credits of $100 on Max or $200 on Ultra. Enterprise sits above Ultra on different pricing logic again, listing a $20 seat and then metering every token at API rates, which the breakdown of Claude Enterprise pricing walks through along with the seat minimums. If that decision is being made for a company rather than for you alone, Claude and ChatGPT compared on enterprise terms sets the seat minimums, SSO and SCIM support, audit logs and retention commitments side by side. If you are weighing the tiers against each other, which Claude plan you actually need compares what each one gates.

Is there a free tier for the Claude API? No

This is the question people get wrong most often, so it deserves a plain answer. The Claude API has no free tier, and the Free and Pro plans come with no monthly API allowance. What it does have, according to Anthropic's pricing documentation as of October 2026, is a small amount of free credits for new users to test the API, and companies evaluating it can ask sales about extended trials. That is a test balance, not an ongoing allowance. The recurring allowance is tied to a paid subscription: as of October 31, 2026, Max, Ultra and Team plans include monthly credits for the Claude Platform, $100 a month on Max and $200 on Ultra, and on Team $20 per Standard seat and $100 per Premium seat, pooled into one balance capped at $500. New subscribers can claim these credits after seven days on an eligible plan, by linking a Console organization. Unused credits expire at the end of each billing cycle, or monthly on annual plans, and Free, Pro and Enterprise plans are not eligible. Without one of those plans, to keep going you create an organization in the Console, add a payment method, buy credits, and every request draws down that balance. How fast that balance drains depends on which model you call, and Fable 5's published rate of $10 per million input tokens and $50 per million output gives you a concrete sense of what a request costs before you buy your first credits.

What the API does have is usage tiers, which are a different thing. Tiers named Start, Build, Scale and Custom set your rate limits, and the first three also carry a monthly spend cap. Your organization is placed on one automatically based on usage history and account standing, and a new organization may begin in an Evaluation tier with lower limits that rise on their own as it builds history. The Start tier carries a $500 monthly spend cap as of October 2026. That cap is a ceiling on what you may spend, not an amount you are given. For the full accounting, including which parts of the platform cost nothing to use and a worked cost for a first small app, the longer answer on whether the Claude API has a free tier runs through the rates in detail. If that first app also needs a database, Supabase's free plan compared with self-hosting it shows that piece can cost nothing, with 500 MB of database and the catch that a project left idle for a week gets paused.

You can see the difference in one request. With a funded key, this returns a message:

curl -s https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{"model":"claude-sonnet-5","max_tokens":16,"messages":[{"role":"user","content":"hi"}]}'

Once the balance is used up, starter credits included, the same request fails with an error about your credit balance instead of returning a message. A wrong or revoked key gives a different error, authentication_error with HTTP 401, so the two failures are easy to tell apart. Add credits on the billing page in the Console, and the same command starts working.

Your subscription and your API key are separate wallets. Paying for Pro does not fund an API key, and API credits do not raise your chat allowance. Max, Ultra and Team are the bridge between them: their monthly Platform credits can be spent by any API key in the linked Console organization, but they expire each billing cycle, or monthly on annual plans, rather than building up. Claude API cost versus a subscription works through which one is cheaper for a given amount of work.

Where paying starts to make sense

Three signals mean the free tier has stopped being the right tool.

• You hit the limit on most days. A limit you meet once a month is a non-event. A limit you meet every afternoon costs you more in waiting than the subscription costs in money.
• You need Claude Code. No amount of free usage unlocks it, so the decision is already made.
• A program is calling Claude, not a person. Anything scripted, scheduled, or bursty belongs on an API key from the start, because the API meters per minute rather than per window, so it does not compete with your own chat allowance.

If you land somewhere between those signals and cannot tell which way you fall, the longer decision guide on whether Claude Pro is worth paying for works the question from how often the free limit actually stops you and where Max, Ultra or an API key fits instead. If you do decide to pay and the checkout then rejects your card, why a Claude plan payment gets declined explains common causes, including an unsupported billing location, and lists every plan price in USD.

If none of those apply, staying free is a reasonable long-term answer and not a compromise. That is where most students land, and what Claude costs a student in 2026 sets out the free tier, the university route, and a month of coursework priced on the API. The decision also runs both ways, and dropping back down to the free plan leaves you on the tier you paid for until that billing period ends, which makes a single month of Pro a cheap way to test those three signals against your own week. Anthropic is also not the only place to put that money, and Claude and ChatGPT plans priced side by side is worth reading before you commit, especially if most of your work is coding. If cost is the whole reason you are asking, there is a third road: run an open model on hardware you rent, which trades a monthly subscription for a server bill you control. running Ollama on a VPS to self-host an LLM covers what that actually takes, including the memory the smaller models need.

FAQ

How many messages can I send to Claude for free?

There is no fixed number, because the free plan meters the cost of your conversation rather than counting messages. A short question costs little and a long thread with attachments costs a lot. Anthropic also states that limits vary with demand. The reliable figure is the reset time shown in the message you get when you run out.

Is Claude free or paid?

Both. The free plan costs $0 and covers chat on the web, mobile and desktop apps, with web search, memory, file creation, Claude Design, Slides and Docs, and up to five projects. Paid plans start at Pro, which as of October 2026 is $17 per month billed annually or $20 billed monthly, and which adds Claude Code, handed-off and scheduled tasks, more models, Research, unlimited projects and a larger allowance. Max costs $100 per month for five times the Pro usage every five hours and every week; Ultra costs $200 per month for ten times the Pro usage every week, with no five-hour limit for verified accounts.

Can I get the Claude API for free?

New users get enough to test it, and eligible researchers can apply for free API credits through Anthropic's External Researcher Access Program. The Claude API has no free tier, but as of October 2026 Anthropic gives new users a small amount of free credits to try it. Beyond that you add a payment method and buy credits in the Console, and requests draw from that balance until it runs out. Max, Ultra and Team subscriptions now include monthly API credits, but those come with a paid plan rather than for free. The usage tiers named Start, Build and Scale are rate limits and monthly spend caps, not free allowances, so being on the Start tier does not mean you have anything to spend.

Does paying for Claude Pro give me API access?

No. A subscription and an API key are billed separately and metered differently. Pro raises your allowance inside the Claude apps, while an API key draws on prepaid credits and is limited per minute by requests and by tokens. If you want both on Pro, you pay for both. Max, Ultra and Team are different: as of October 31, 2026 they include monthly Claude Platform credits, with $100 a month on Max and $200 on Ultra.

What resets my Claude free tier limit?

The five-hour session allowance refills when its window ends, and the weekly allowance resets at its assigned weekly time. The interface shows your reset time when it stops you. Anthropic also occasionally offers limit resets to eligible plans; if your account has one, Settings > Usage shows a "Reset for free" button. Starting a new chat, switching model, signing out, or reinstalling the app does not restore it, because the limit belongs to your account and not to the conversation.