Guides › New to AI

Get more out of your Claude plan for less

Updated Oct 22, 2026. Based on the topics in Anthropic's own workshop, Maximizing value from Claude Code¹⁷, with every fact checked against Anthropic's help pages and docs (where they differ, we follow the docs).

Running out of Claude halfway through the day? Most of the time it isn't your plan, it's how the conversation is set up. These habits make the same plan go further, in the Claude app and in Claude Code.

What actually uses up your limit

Every time you send a message, Claude reads the whole conversation again, plus any files, instructions and tool results in it. That's measured in tokens (small chunks of text). Anthropic lists what affects your usage: message length, attachment size, conversation length, tools like web search, which model you use, the effort setting, artifacts and multi-step tasks. ¹

• Two limits: a 5-hour session limit and, on paid plans, a weekly limit. ^4,5
• One shared pool: Claude on the web, desktop, phone and Claude Code all count toward the same limits. ^4,6
• Check where you are: Settings › Usage shows your progress toward both limits. ¹

1. Start a fresh chat for each new topic

A chat that wandered from your tax question to a recipe to a work email makes Claude re-read all three every time you reply. Anthropic's tips for making usage last include starting new conversations for new topics. ²

Simple test: if your next question would make sense to someone who hadn't read the chat so far, open a new chat.

2. Attach the file itself, once, in the right place

• Attach only what's needed. Attachment size counts toward usage. If you only need page 12, say so, or attach just that section. ¹
• Reusing the same documents? Put them in a Project. Files you upload to a Project are cached, and cached content counts less against your limits than new content. Anthropic's example: put a research paper's reference materials in one Project and ask lots of questions about them. ¹
• Big Projects are fine on paid plans. When Project knowledge gets close to the limit, Claude switches on a search mode (RAG) to fit more. ⁷
• Batch your questions. Ask three related things in one message instead of three separate messages. ¹

3. Model vs effort: what to change when Claude misses

Claude has two dials, both next to the send button: the model (Haiku, Sonnet, Opus and so on) and the effort level, which controls how hard it thinks before answering. ⁸

⁸

• Higher effort uses more tokens and hits limits faster. Changes apply from the next reply. ⁸
• Try effort before a bigger model. Anthropic: "Tuning effort is often a better lever than switching models." ³
• When to switch model: if Claude keeps missing what the task needs (not just rushing), try a stronger model for that task, then switch back. Anthropic's Claude Code advice is Sonnet for most work, Opus when you're genuinely stuck, and Haiku for quick mechanical jobs. ⁹
• First, check your question. A missed answer is often missing context. Say who it's for, what good looks like, and give an example.

More on picking: Which AI model and effort setting should I use?

4. Summarize and restart long chats

When a chat is long but you still need what's in it:

1. Ask: "Summarize everything we've decided so far, including key facts, numbers and next steps, in under 300 words."
2. Copy the summary.
3. Start a new chat (inside the same Project if you use one) and paste it in with your next question.

You keep what matters and drop the clutter Claude was re-reading. On paid plans you can also ask Claude to search your past chats instead of pasting things again. ^10,1

5. Use Projects and memory so you stop repeating yourself

• Project instructions apply to every chat in that Project ("You're helping me run a small online shop. Keep answers short."). ⁷
• Chats in a Project don't see each other unless the information is added to the Project's knowledge. ⁷
• Memory is on by default on Free, Pro and Max. Claude saves short notes about your role, preferences and ongoing work, and each Project keeps its own memory. You can say "remember this", or turn memory off for a single chat from the + menu. ¹⁰
• Free plans can have up to 5 Projects. ⁵

6. Mind the breaks, and batch your work

Claude keeps recent content in a short-term cache so re-reading it is cheaper. Anthropic says caches expire after a period of inactivity, so if you come back to a Project after a long break, your first message counts that content in full again. ¹ If you can, do related work in one sitting rather than in dribs and drabs across the day.

7. Choosing the right Claude plan

Prices from our Claude plan data (official USD and CAD, never converted). ^5,11

• Pro gives at least 5 times Free's usage; Max gives 5 or 20 times Pro's. ^5,11
• Before upgrading, try usage credits. Pro and Max can keep going past a limit at API rates, with a monthly spend limit you set. That can be cheaper than upgrading if you only go over now and then. ^2,12
• Subscribe on the website. Anthropic's prices above are for web subscriptions. ¹¹

For Claude Code users: the commands that save the most

New to Claude Code? Start with Claude Code explained: what it is and how to set it up.

Claude Code shares your plan's limits with chat, so the same habits apply. Anthropic's own guide to stretching your usage is Models, usage, and limits in Claude Code⁹, and the full reference is Manage costs effectively¹².

^12,9,13

• Keep CLAUDE.md under about 200 lines. It's read on every turn. Move instructions you only need sometimes into skills, which load on demand. ^12,14
• Point at files by path instead of pasting them, and trim logs to the relevant lines. ⁹
• Send noisy work to a subagent (running tests, reading long logs) so only the summary comes back. It still counts toward your limits. ^12,15
• Breaks and the cache: on a Claude subscription, Claude Code keeps the main conversation's cache for an hour within your included usage (five minutes once you're on usage credits). After a longer break the next message re-reads everything at full price, and /compact costs the most when you resume an old session. /clear costs nothing. ^13,12
• Agent teams are experimental, off by default, and use far more tokens. Keep them small and use Sonnet for teammates. ^16,12
• Teams: Team and Enterprise admins can set spend limits for the organization or for individual people. ¹²

Five minutes or an hour? You'll often hear that Claude's cache lasts five minutes. That's the standard API setting. Anthropic's current Claude Code docs say a subscription gets a one-hour cache for your main conversation within your included usage, so we follow the docs. ¹³

Common questions

5 quick questions, answered

Why do I hit my Claude limit so fast?

Usually because of long conversations. Claude re-reads the whole conversation every time you send a message, so a long chat costs more with each reply. Big attachments, tools like web search, the Opus model and higher effort settings use up limits faster too. Starting a new chat for each new topic is the biggest single fix.

When do Claude usage limits reset?

Claude uses a 5-hour session limit plus a weekly limit on paid plans. Anthropic's pricing page describes the session limit as a rolling 5-hour window. You can see both, and your next reset time, in Settings > Usage.

Should I switch to Opus or raise the effort?

Try effort first for hard thinking problems. Anthropic says tuning effort is often a better lever than switching models. Switch to a stronger model when Claude keeps missing what a task needs, and drop back afterwards.

Do Claude chat and Claude Code share the same limit?

Yes. Anthropic says usage across Claude on the web, desktop, mobile and Claude Code all counts toward the same limits.

Is Claude Max worth it over Pro?

Only if you regularly hit Pro's limits after trying the tips in this guide. Max gives five times Pro's usage every 5 hours and every week. Ultra gives ten times Pro's weekly usage, with no 5-hour limit once your account is verified. Anthropic's own advice to Pro users who keep hitting limits is to consider Max (called Max 5x until Oct 30, 2026).

The AI agents series

• AI agents, explained for normal people
• Get more out of your Claude plan for less
• How to run a small team of AI agents (no coding)
• What Grok Bot can actually do for you
• AI agents compared: prices and features

Related: Claude plans and prices · Get the most out of Claude · Claude Max API credits · Claude Opus 5.5

New to AI?

AI crash course: start here

What AI chatbots are, how to ask, and how to stay safe. About 10 minutes.

Should I pay for AI?

What free plans really give you, what a paid plan adds, and who needs one.

Which AI model should I use?

Pick the right model and thinking level in 60 seconds.

Tip: subscribe on the website, not in the app

Buying through the iPhone or iPad app often costs more because of Apple's App Store fees. Sign up on the provider's own site (like claude.ai or chatgpt.com) and you can still use the app with the same account. How to switch and what to know · Compare app and website prices.