Somebody typed “Good day Claude” and used 13% of their session restrict.
That is an actual Reddit put up from an actual one that opened Claude, despatched a greeting, and watched greater than one-eighth of their utilization disappear earlier than asking a single query.

A separate person on X reported ending up in a “four-hour cooldown jail” from the identical set off. The factor is, no one had clarification for why it occurred.
The reply is tokens. Most individuals utilizing LLMs right this moment don’t have any framework for understanding what a token is, why it prices what it prices, or the place their utilization goes earlier than they’ve completed something helpful. Each main LLM – Claude, GPT-5, Gemini, Grok, Llama and so forth. runs on the identical underlying economics. Tokens are the forex of this whole business.
For those who use any of them repeatedly, understanding how tokens work is the distinction between getting actual work completed and hitting your restrict at 11am.
Let’s decode.
What a Token Really Is
Consider a token as a bit of textual content someplace between a syllable and a phrase in dimension.
Let’s simply say “Improbable” is one token. “I’m” is 2 tokens. “Unbelievable” is perhaps three tokens relying on the mannequin, as a result of some fashions break unfamiliar or lengthy phrases into subword items. The OpenAI tokenizer playground (platform.openai.com/tokenizer) permits you to paste any textual content and see precisely the way it will get chopped up in coloured blocks. Value attempting as soon as simply to calibrate your instinct.

The tough conversion for English: 1,000 tokens ≈ 750 phrases ≈ 2-3 pages of textual content. One token averages about 4 characters or 0.7 phrases. A typical 800 phrase weblog put up is roughly 1,000-1,100 tokens.
These numbers solely maintain for English. Code tokenization is worse: 1.5 to 2.0 tokens per phrase, as a result of programming syntax has plenty of characters that do not map cleanly onto pure language tokens. Chinese language, Japanese, and Korean are worse nonetheless, consuming 2 to eight instances extra tokens than English for equal content material. For those who write plenty of code or work in a non English language, your consumption is meaningfully increased than the back-of-envelope math suggests.
Totally different fashions use completely different tokenizers, so the identical textual content does not price the identical tokens all over the place. 1,000 tokens on GPT-5 (which makes use of the o200k_base tokenizer) is perhaps 1,200 tokens on Claude or 900 tokens on Gemini. Evaluating utilization throughout platforms requires utilizing every mannequin’s particular tokenizer for correct counts.
The Context Window
Tokens are necessary for 2 distinct causes. The primary is your utilization restrict: how a lot you are able to do earlier than hitting a wall. The second is the context window: how a lot the mannequin can maintain in reminiscence directly.
Each mannequin has a context window measured in tokens. Claude Sonnet 4.6 helps 1 million tokens. GPT-5 has 400K. Gemini 3 Professional has 2 million. Llama 4 Scout has 10 million. These numbers are spectacular however deceptive.
Bigger context home windows do not routinely imply higher efficiency. Analysis constantly reveals fashions degrade in high quality earlier than reaching their said limits. A 2024 examine from researchers Levy, Jacoby, and Goldberg discovered that LLM reasoning efficiency begins degrading round 3,000 tokens, properly earlier than any mannequin’s technical most. A 2025 examine from Chroma examined 18 fashions together with GPT-4.1, Claude 4, and Gemini 2.5 and documented what they referred to as “context rot”: a progressive decay in accuracy as prompts develop longer, even on easy string-repetition duties. Each mannequin confirmed that extra context shouldn’t be all the time higher.
The context window can also be shared by every part, not simply your message and the mannequin’s reply. System directions, software calls, each earlier flip within the dialog, uploaded recordsdata, and inner reasoning steps all eat from the identical pool.

The Six Silent Token Drains
The bulk assume token utilization seems like: I kind one thing, the mannequin responds, that is one alternate. However in actuality, it’s not linear and predictable.
1. Dialog Historical past Compounds Quick
Each message you ship in a multi-turn dialog carries all the prior dialog as context. Flip 1 prices 2 items: you ship 1, the mannequin sends 1 again. Flip 2 prices 4 complete as a result of your second message consists of the primary alternate. Flip 3 prices 6. By flip 10, you may need spent 110 items cumulatively. Those self same ten duties as ten separate one-turn conversations would price 20 items complete. Similar output however 5 and a half instances inexpensive.
Individuals who deal with a dialog like a working doc, including to the identical thread for hours as a result of it feels organized, are doing probably the most token-expensive factor doable.
A concrete instance: you are utilizing Claude to debug a software program mission. You paste 2,000 tokens of code, ask a query, get a solution, ask a follow-up, and so forth. By the fourth alternate, the mannequin is processing roughly 12,000 tokens to reply a query that, in isolation, would price 500. The amassed historical past is doing many of the spending.
2. Prolonged Pondering Generates Tokens You By no means See
Most main LLMs now have a reasoning mode. OpenAI calls it o-series. Google calls it Pondering Mode. Anthropic calls it Prolonged Pondering. When enabled, the mannequin works by the issue internally earlier than responding.
That inner reasoning generates tokens. Reasoning tokens can quantity to 10 to 30 instances greater than the seen output. A response that appears like 200 phrases to you may need price 3,000 reasoning tokens behind it.
Claude’s Prolonged Pondering is now adaptive, which means the mannequin decides whether or not a activity wants deep reasoning or a fast reply. On the default effort stage, it nearly all the time thinks. So whenever you ask Claude to repair a typo, reformat an inventory, or search for a primary truth, it is nonetheless burning pondering tokens on an issue that does not require them. Toggling Prolonged Pondering off for easy duties reduces prices with no high quality tradeoff.
The identical situation applies to OpenAI’s reasoning fashions. GPT-5 routes requests to completely different underlying fashions relying on what your immediate indicators. Phrases like “assume exhausting about this” set off a heavier reasoning mannequin even when you do not want one. OpenAI’s personal documentation warns in opposition to including “assume step-by-step” to prompts despatched to reasoning fashions, because the mannequin is already doing it internally.
3. System Prompts Run on Each Request
Any AI product constructed on a basis mannequin, together with customized GPTs, Claude Initiatives with customized directions, or enterprise deployments, prepends a system immediate to each message you ship.
A typical system immediate runs 500 to three,500 tokens. Each time you ship something, these tokens run first. An organization working an inner chatbot with a 3,000-token system immediate dealing with 10,000 messages per day spends 30 million tokens on directions alone, earlier than any person has requested something significant.
On the particular person stage: a Claude Mission with in depth customized directions reruns these directions each time you open the mission. Protecting mission information tight is immediately cheaper, not simply neater.
4. The “Good day” Downside
Again to the Reddit put up. How does “hi there” eat 13% of a session?
Really earlier than processing your phrase “hi there”, it masses the system immediate, mission information, dialog historical past from earlier within the session, and enabled instruments. In Claude Code particularly, it masses CLAUDE.md recordsdata, MCP server definitions, and session state from the working listing. All of that’s billed as enter tokens on each alternate, together with the primary one.

In case your Claude Code atmosphere has a fancy CLAUDE.md, a number of MCP servers enabled, and a big mission listing, your baseline token price per message earlier than you have typed something may already be a number of thousand tokens. And “Good day” in that atmosphere prices one phrase plus all of the infrastructure the mannequin must load earlier than it will probably reply.
5. Uploaded Information Sit on the Meter Constantly
Importing a 50-page PDF to a Claude Mission signifies that doc is held in context even whenever you’re not actively asking questions on it. It consumes tokens each session as a result of the mannequin wants consciousness of it to reference it when wanted.
Token consumption in any chat comes from uploaded recordsdata, mission information recordsdata, customized directions, message historical past, system prompts, and enabled instruments, on each alternate. For those who add 5 giant paperwork you ended up not referencing, you are still paying for them.
Preserve mission information matched to what you are really engaged on. Deal with it like RAM, not a submitting cupboard.
6. Agentic Instrument Calls Explode the Depend
For those who use AI brokers, Claude with instruments, ChatGPT with Actions, or any autonomous workflow the place the mannequin calls exterior APIs or searches the net: each software name appends its full outcome to the context. An online search returns roughly 2,000 tokens of outcomes. Run 20 software calls in a single session and you have consumed round 40,000 tokens in software responses alone, earlier than factoring within the rising dialog historical past stacking on prime.
Claude Code brokers performing 10 reasoning steps throughout a big codebase can course of 50,000 to 100,000 tokens per activity. For a workforce of engineers every working a number of agent classes per day, this turns into the first price driver.
Easy methods to Protect Your Token Finances
Begin a New Dialog for Each New Process
Given the compounding math above, preserving one lengthy dialog open throughout a number of unrelated duties is the most costly method to make use of an LLM. A ten-turn dialog spanning 5 subjects prices greater than 5 2-turn conversations protecting the identical floor.
The intuition to maintain every part in a single thread feels organized. However resist it. So comply with: new activity, new dialog.
Match the Mannequin to the Work
Frontier fashions, Claude Opus, GPT-5, and Gemini 3 Professional, are dearer than their smaller siblings, and for many duties the standard distinction is negligible. Claude Sonnet handles complicated coding, detailed evaluation, long-form writing, and analysis synthesis with out significant high quality loss versus Opus. The distinction reveals up solely on severely complicated multi-step reasoning, which represents a fraction of precise day by day utilization.
Default to the mid-tier mannequin (Sonnet, GPT-4o, Gemini Flash Professional). Use the flagship when the duty genuinely calls for it. Keep away from this:

Flip Off Prolonged Pondering for Easy Duties
For Claude: toggle Prolonged Pondering off underneath “Search and instruments” when doing fast edits, brainstorming, factual lookups, or reformatting. Response high quality on these duties will not change. Token price drops considerably.
For GPT: use customary GPT-4o fairly than o-series fashions for something that does not require deep multi-step reasoning. The o-series is purpose-built for exhausting reasoning issues and wasteful for every part else.
Write Shorter Prompts
The analysis says quick prompts usually work higher than lengthy ones, they usually’re cheaper. The sensible candy spot for many duties is 150-300 phrases. That is particular sufficient to offer the mannequin actual path with out stuffing it with context it does not want.
Write the shortest model of your immediate that describes your intent. Check it. Add solely what’s really lacking within the output.
For instance, as a substitute of: “I am engaged on a advertising marketing campaign for a B2B SaaS product that helps finance groups automate their accounts payable workflows. I might such as you to assist me write a topic line for an electronic mail going to CFOs at mid-market corporations. The tone must be skilled however not overly formal. It ought to convey urgency with out being pushy. The e-mail is a part of a drip sequence and that is the third electronic mail within the collection, which implies the recipient has already heard from us twice and hasn’t responded but…”
Strive: “Write 5 topic traces for electronic mail #3 in a B2B drip to CFO prospects. Product: AP automation SaaS. Tone: skilled, slight urgency.”
The output is similar high quality. The token price is a fraction.
Skip Pleasantries Inside Periods
Each “thanks, that is useful!” or “nice, now are you able to additionally…” extends the dialog and inflates the working context. In a token-constrained atmosphere, social filler prices actual utilization for no informational profit.
That is additionally the mechanical clarification for the “hi there” downside. In a loaded atmosphere, a greeting is a full flip that masses all of the infrastructure and generates a full response for zero informational worth. Mixed with a fancy system atmosphere, that provides as much as 5-10% of a session earlier than any actual work begins. And that is cap:

Request Structured Outputs
Asking for structured outputs, resembling JSON, numbered lists, or tables, usually requires fewer output tokens than narrative explanations whereas producing extra usable outcomes. Specifying “Checklist 3 product options as JSON with keys: characteristic, profit, precedence” generates a parseable response in fewer tokens than “describe the three most necessary product options intimately.”
Analysis on this sample reveals output token reductions of 30-50% for equal informational content material.
Preserve Mission Information Matched to the Present Process
Solely embrace paperwork immediately related to what you are engaged on now. Archive previous recordsdata when a mission part ends. Each file in a Claude Mission runs on each session whether or not you reference it or not.
Easy methods to Test What You Have Left
Most AI merchandise do not present a token meter. Here is how you can discover your utilization anyway, by platform.
Claude (claude.ai)
Go to Settings → Utilization, or navigate on to claude.ai/settings/utilization. This reveals cumulative utilization in opposition to your plan’s restrict. It is a lagging indicator and does not present real-time token rely inside a dialog.

For Claude Code particularly: /price reveals API-level customers their token spend for the present session damaged down by class. /stats reveals subscribers their utilization patterns over time.

Third-party instruments for Claude Code
ccusage is a CLI software that reads Claude’s native JSONL log recordsdata and reveals utilization damaged down by date, session, or mission. It runs as a one-line npx command with no full set up. For Professional and Max subscribers who cannot see consumption within the Anthropic Console (as a result of they pay a flat subscription fairly than per-token), that is the first option to observe the place utilization goes.
Claude-Code-Utilization-Monitor gives a real-time terminal UI with progress bars, burn charge analytics, and predictions for when your present session will run out. It auto-detects your plan and applies the correct limits: Professional is round 44,000 tokens per 5-hour window, Max5 round 88,000, and Max20 round 220,000. Run it in a separate terminal window and you may see consumption replace dwell.
Claude Utilization Tracker is a Chrome extension that estimates token consumption immediately within the claude.ai interface, monitoring recordsdata, mission information, historical past, and instruments, with a notification when your restrict resets.
ChatGPT
OpenAI does not expose token utilization to client customers immediately. Developer accounts with API entry can see per-request token counts at platform.openai.com/utilization. Client subscribers don’t have any native meter. Third-party extensions exist within the Chrome retailer however aren’t formally supported.
API customers (any platform)
Each API response consists of token counts within the metadata. For Claude, input_tokens and output_tokens seem in each response object. For OpenAI, the equal fields are utilization.prompt_tokens and utilization.completion_tokens. Construct logging round these fields from the beginning, it is the one dependable option to observe consumption at scale.
Earlier than you ship: token counters
Instruments like runcell.dev/software/token-counter and langcopilot.com/instruments/token-calculator allow you to paste textual content and get an prompt rely earlier than sending, utilizing every mannequin’s official tokenizer. No signup are required and it runs within the browser. Helpful earlier than submitting giant paperwork or complicated prompts.
The Talent Value Having
Token literacy was once a developer concern however not right this moment.
The identical shift occurred with knowledge. Ten years in the past, knowledge literacy meant SQL and spreadsheets, practitioner territory. Now each enterprise decision-maker is anticipated to learn a dashboard, interpret a funnel, and query a metric. Tokens are on the identical trajectory.
LLMs are embedded in actual work now: drafting, evaluation, coding, analysis. The individuals who perceive the underlying economics will use them extra successfully, hit limits much less usually, and get extra from the identical subscription.
Cheers.

