Claude free tier: what the limits actually are

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 July 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, and desktop extensions. It does not include Claude Code or Claude Cowork, which start on Pro.

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 window rather than by a fixed lifetime quota, so it refills on its own and you can come back later the same day.

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 resends the whole conversation to the model, so a long chat costs more per message than a fresh one. The tenth message in a thread is more expensive than the first, which is why a long thread hits 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. The general mechanics of session and weekly windows are the same on paid plans, and 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 costs you the whole history again 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 July 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.

• Claude Code and Claude Cowork. These are the agentic surfaces, and they are gated behind Pro. If your reason for using Claude is to have it work in a terminal or on a repository, the free plan cannot do it at any usage level.
• Model choice. Pro gives you more than one model to pick from, which matters when you want a cheaper model for routine work and a stronger one for hard problems.
• Unlimited projects and Research. Free covers chat. Pro covers the workflow around it.
• More usage. The allowance is larger, and it is the same rolling system, so it also resets rather than expiring for the month.

Above Pro, Max starts at $100 per month and is sold as 5x or 20x the Pro usage, with priority access during peak hours. That is a capacity purchase, not a feature purchase. 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 no free monthly allowance. You create an organization in the Console, add a payment method, buy credits, and every request draws down that balance. There is no trial pool of tokens that arrives with the account.

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 your monthly spend cap, and your organization is placed on one automatically based on usage history and account standing. The Start tier carries a $500 monthly spend cap as of July 2026. That cap is a ceiling on what you may spend, not an amount you are given.

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"}]}'

With an unfunded organization the same request fails, because the balance is checked before the model runs. The response carries an error whose message says your credit balance is too low to access the API. 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. 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 none of those apply, staying free is a reasonable long-term answer and not a compromise. 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, so the same allowance can be ten messages or many more. 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 and file creation. Paid plans start at Pro, which as of July 2026 is $17 per month billed annually or $20 billed monthly, and which adds Claude Code, Claude Cowork, model choice, unlimited projects and a larger allowance. Max starts at $100 per month for 5x or 20x the Pro usage.

Can I get the Claude API for free?

No. The Claude API has no free tier and no starter credits. You add a payment method and buy credits in the Console, and requests draw from that balance until it runs out. 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, you pay for both.

What resets my Claude free tier limit?

Time, and nothing else. The allowance refills on its own when the current window ends, and the interface names that time when it stops you. 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.