There’s a particular frustration that hits when you run out of Claude tokens mid-project—and Anthropic is more than happy to sell you a $100 or $200 plan to solve that.But chances are, the problem isn’t that you use Claude more—it’s that you use it inefficiently.I’ve been on the $20 Pro plan since 2024 and have never once burned through my weekly limit.
Here’s exactly what I do differently.Actively monitor your token usage You can't improve what you don't measure Most AI chat apps give you a fixed message limit where you get many prompts per day, and that's it.But Claude works differently.
It gives you a token limit, and how quickly you burn through that limit depends entirely on what you're asking it to do.Tokens are the small chunks of text AI models process internally.As a rough estimate, 100 English words translate to around 133 tokens, though some languages consume more than others.
That means short prompts with concise responses can let you send hundreds of messages before you get anywhere near your limit.But if you paste in a 10,000-word brief and ask Claude to build an app from scratch, that single conversation can burn through a significant portion of your token budget.Once you understand this, it becomes much easier to manage your usage proactively.
To monitor your token usage, open the Claude website or mobile app and head to Settings > Usage.It’ll show you your weekly limit and current session usage (I'll explain what a "session" means in a later section).If you're using Claude Code or Cowork—which if you're not, you should—typing /context shows exactly how many tokens the current chat has consumed.
What the math actually looks like I had a conversation with Claude where we discussed my daily routine and built an HTML artifact to help me visualize it.The entire chat was only around 10–15 messages, but it consumed roughly 63,500 tokens.Before the conversation, my session usage sat at 2% and my weekly usage at 10%.
Afterward, the session jumped to 17%, while the weekly limit only increased to 12%.In other words, roughly 15% of a session translated to about 2% of my weekly quota.Extrapolating from that, a completely full session appears to equal around 13% of your total weekly allowance.
Practically speaking, that means you get roughly 7–8 heavy working sessions per week.You can burn through them all in a weekend if you want, or spread them more evenly across the workweek.Spread your work across the day Strategically use your Claude session Claude has two separate usage limits: a session limit and a weekly limit.
A session lasts roughly five hours.The timer starts the moment you send your first message, then automatically resets five hours later.The countdown won’t start again until you send a new message.
That means if you're working on a large project and hit your session limit midway through, you don't necessarily need a more expensive plan—you may just need to manage your sessions more strategically.For example, you can intentionally trigger the session timer early by sending a simple message like "Hi" a few hours before you actually plan to work.Then, once the session is close to resetting, you can sit down and begin your real project work.
The advantage is that even if you burn through your limit during that work session, the next session window opens shortly afterward.In practice, this can let you use the equivalent of two Claude sessions in a single 2–3 hour focused work block.Pro Tip: Schedule tasks to work across session resets If you're using Claude Code or Cowork and your files are saved locally, you can schedule tasks that automatically resume work once the next session window opens.
This means that even if you're away from your computer when your limits reset, Claude can continue working on the project automatically.Just create a scheduled task at the time your session restarts, where you tell it the goal, current progress, and point it to the file (or collection of files) you’ve been working on.Cowork also has a Dispatch feature that lets you control your desktop session remotely from the Claude mobile app.
So if a new session opens while you're out, you can kick off the next stage of work directly from your phone instead of needing to return to your desk.Related I let Claude take over my PC—and it automated these 5 boring tasks for me Claude can do a lot more than answer your questions—it can automate the boring stuff, so you can focus on the fun stuff.Posts By Dibakar Ghosh Keep your conversations short and focused Token consumption compounds the longer you chat Here's something most people don't realize: Claude rereads the entire conversation every time you send a new message.
So if your first exchange (your message plus Claude’s response) used 200 tokens, and your next message adds another 100, Claude isn't just processing the 100 tokens from the new message—it’s processing all 300 tokens together.So, by the time you're 15 or 20 messages deep, the accumulated context starts eating into your token budget surprisingly quickly.Fortunately, the problem has a simple fix.
Just start a fresh chat every 15–20 messages.That might sound disruptive at first, but in practice it works surprisingly well because Claude can search your previous conversations.You can simply open a new chat and say something like: Continue where we left off in my previous chat about X.
Claude will pull in the relevant context without carrying over the full weight of the original thread.This way you keep the continuity while significantly reducing token overhead.I should also mention that if you're using Claude primarily for coding and don't care about conversational niceties, there's a Claude Code skill called Caveman that forces Claude to respond in stripped-down, minimal language.
No more "Great question!" or "I'd be happy to help." According to the project's creator, it can reportedly reduce token usage by around 65%.Pro Tip: Edit your messages instead of correcting Claude When Claude misunderstands you, the natural instinct is to send a follow-up message like: “Actually, I meant...” or “No, I need it to do X instead.” Don’t do that.Just edit your original message instead.
Whenever you send a correction as a message, both the mistake and the correction become part of the conversation history that Claude continues processing with every future message.As we just discussed, this can eat into your token limits.However, when you edit the original message, the incorrect branch of the conversation effectively disappears, which helps keep the thread cleaner and reduces unnecessary context accumulation.
Just note that this only works in Claude.ai on the web and mobile app.Claude Code and Cowork currently don't support editing previously sent prompts.Related 5 Useful AI Prompting Techniques You Should Know Level up your AI game with these pro prompting strategies.
Posts By Dibakar Ghosh Stop using Opus for everything Sonnet and Haiku are more capable than people give them credit for Claude gives you access to three models: Haiku, Sonnet, and Opus.Since Opus is the most capable model, many people simply select it by default and never switch away from it.The instinct makes sense.
If you're paying for access to the "best" model, why settle for something less capable? But in practice, most everyday tasks don't actually need Opus-level reasoning.Sonnet is already more than capable of handling everyday work like emails, general research, brainstorming, summaries, and even lightweight coding tasks.And for simpler administrative work—organizing files, rewriting text, updating to-do lists, or formatting content—Haiku is often perfectly sufficient.
Using Opus for these smaller workloads is usually unnecessary overkill.This matters because Opus is significantly more expensive.A conversation that might consume 10% of your session on Sonnet could easily consume 25–30% on Opus instead.
And because Claude continuously rereads conversation history, that efficiency gap becomes even larger in long chats.Personally, I default to Sonnet for almost everything.I only switch to Opus when I notice Sonnet failing or struggling with deeper reasoning, complex planning, or difficult coding problems.
Don't load all your connectors in every chat Every enabled connector costs tokens even before you've typed a word Every app connector (MCP server) you enable adds overhead to your token usage before you've typed a single message.When a new chat starts, Claude loads instructions and context for each connected service—Google Drive, Notion, Gmail, and anything else you have enabled.Even if you never end up using those tools during the conversation, that setup context still contributes to your token consumption.
Because of that, it's worth disabling connectors you don't actually need for the task at hand.For example, if you're working on a vibe coding project that doesn’t need access to your Notion or Google Calendar, turning those connectors off before starting the chat can help reduce unnecessary token overhead.It only takes a few seconds, but over time, it can noticeably improve how efficiently you use your quota.
Related I transformed Claude into the ultimate coding tutor that tracks my progress (Prompt included) I built a coding tutor that won't let me cheat my way through it.Here's the prompt.Posts By Dibakar Ghosh Consider buying a second $20 account instead of jumping to Max The gap between Pro and Max is bigger than it needs to be Even after implementing all these strategies, if the $20 Pro plan still feels limiting, don’t upgrade to the $100 (or $200) Max plan just yet.
Yes, it’s good value where you’re getting 5x (or 20x) the usage limit—but will you use that extra limit for which you’re so generously paying for? For many people, the answer is probably no.You may only need 2× or 3× more capacity than the standard Pro plan provides.And if that's the case, buying a second—or even third—$20 Pro account can often be the more economical option.
You spend less overall while still getting a much more comfortable token budget for your workload.Some people worry this means losing all the context and personalization they've built up over time on their primary account.In practice, though, that usually isn't a major issue.
Claude supports importing memory from other chatbots, and you can configure the same connectors, skills, and MCP servers across accounts to recreate most of your working environment.If you're using Claude Code or Cowork, the setup becomes even easier because multiple accounts can point to the same working directory.That means all of them can access the same project files and context—the only thing you're really switching between is the active account.
Related I use Claude, Gemini, and ChatGPT every day—here's the only one you should pay for The battle of the bots.Posts 18 By Adam Davidson These habits work for Max subscribers too These are the strategies I personally use to make the most of the $20 Claude Pro plan.I use Claude daily for both work and personal projects, and I've never once exhausted my weekly token limit.
That said, if you're consistently doing large-scale coding, research, or reasoning tasks that genuinely require Opus-level capability for long periods of time, then upgrading to Max can absolutely make sense.But even then, the habits in this article still matter.Efficient prompting, shorter conversations, strategic model selection, and smarter session management will stretch a Max subscription just as effectively as they stretch a Pro one.
Claude Price $20 Claude is an AI assistant made by Anthropic. It can assist with a wide range of tasks—writing, coding, analysis, research, and more. Unlike a search engine, Claude reasons through problems conversationally, making it useful as a thinking partner rather than just an information retrieval tool.See at Claude Expand Collapse
Read More
