---
title: Why Claude's Token Cost Is Higher Than Ever, by Design
description: The Claude token cost per task keeps climbing while the rate card barely moves. Default thinking, tokenizer changes and quiet cache cuts explain the bill.
canonical: https://lukeocodes.dev/claude-token-cost
type: article
locale: en_GB
published: 2026-08-22T00:00:00.000Z
modified: 2026-08-22T00:00:00.000Z
authors:
  - Luke Oliff
section: Opinion
tags:
  - Opinion
  - AI
  - Pricing
  - anthropic
  - claude
  - tokens
  - api pricing
feed: https://lukeocodes.dev/rss.xml
search: https://lukeocodes.dev/search.json?q={query}
schema:
  - "@context": https://schema.org
    "@graph":
      - "@type": WebSite
        "@id": https://lukeocodes.dev/#website
        url: https://lukeocodes.dev
        name: Luke Oliff
        alternateName: lukeocodes
        inLanguage: en
        publisher:
          "@id": https://lukeocodes.dev/authors/luke#person
        description: Personal blog and writing by Luke Oliff — developer relations, voice AI, developer experience.
        potentialAction:
          - "@type": SearchAction
            target:
              "@type": EntryPoint
              urlTemplate: https://lukeocodes.dev/blog/search?q={search_term_string}
            query-input: required name=search_term_string
      - "@type": Person
        "@id": https://lukeocodes.dev/authors/luke#person
        name: Luke Oliff
        url: https://lukeocodes.dev/authors/luke
        description: Engineer and developer relations leader. Writing about developer relations, AI, and the voice industry, from the UK.
        mainEntityOfPage: https://lukeocodes.dev/authors/luke
        givenName: Luke
        familyName: Oliff
        image: https://lukeocodes.dev/authors/luke.jpg
        jobTitle: Engineer and developer relations leader
        sameAs:
          - https://www.linkedin.com/in/lukeocodes
          - https://github.com/lukeocodes
          - https://bsky.app/profile/lukeocodes.dev
          - https://x.com/lukeocodes
          - https://hashnode.com/@lukeocodes
          - https://dev.to/lukeocodes
        knowsAbout:
          - Developer Relations
          - Developer Experience
          - Voice AI
          - Text-to-Speech
          - Answer Engine Optimisation
          - Search Engine Optimisation
          - Open Source
          - Developer Tooling
          - API Design
          - Speech-to-Text
          - Voice Agents
          - Community Building
        knowsLanguage:
          - English
        nationality: British
        alumniOf:
          - "@type": Organization
            name: The Open University
  - "@context": https://schema.org
    "@graph":
      - "@type": Person
        "@id": https://lukeocodes.dev/authors/luke#person
        name: Luke Oliff
        url: https://lukeocodes.dev/authors/luke
        description: Engineer and developer relations leader. Writing about developer relations, AI, and the voice industry, from the UK.
        mainEntityOfPage: https://lukeocodes.dev/authors/luke
        givenName: Luke
        familyName: Oliff
        image: https://lukeocodes.dev/authors/luke.jpg
        jobTitle: Engineer and developer relations leader
        sameAs:
          - https://www.linkedin.com/in/lukeocodes
          - https://github.com/lukeocodes
          - https://bsky.app/profile/lukeocodes.dev
          - https://x.com/lukeocodes
          - https://hashnode.com/@lukeocodes
          - https://dev.to/lukeocodes
        knowsAbout:
          - Developer Relations
          - Developer Experience
          - Voice AI
          - Text-to-Speech
          - Answer Engine Optimisation
          - Search Engine Optimisation
          - Open Source
          - Developer Tooling
          - API Design
          - Speech-to-Text
          - Voice Agents
          - Community Building
        knowsLanguage:
          - English
        nationality: British
        alumniOf:
          - "@type": Organization
            name: The Open University
      - "@type": BlogPosting
        "@id": https://lukeocodes.dev/claude-token-cost#article
        headline: Why Claude's Token Cost Is Higher Than Ever, by Design
        alternativeHeadline: Why Claude's Token Cost Is Higher Than Ever, by Design
        description: The Claude token cost per task keeps climbing while the rate card barely moves. Default thinking, tokenizer changes and quiet cache cuts explain the bill.
        datePublished: 2026-08-22T00:00:00.000Z
        dateModified: 2026-08-22T00:00:00.000Z
        url: https://lukeocodes.dev/claude-token-cost
        mainEntityOfPage: https://lukeocodes.dev/claude-token-cost
        inLanguage: en
        author:
          - "@id": https://lukeocodes.dev/authors/luke#person
        isPartOf:
          "@id": https://lukeocodes.dev/#website
        keywords: Opinion, AI, Pricing, anthropic, claude, tokens, api pricing
        articleSection: Opinion
        about:
          - "@type": Organization
            name: Anthropic
        abstract: Opus 4.5 was the efficient one. Everything since burns more tokens per task with thinking on by default, a costlier tokenizer, and a prompt cache TTL that got quietly cut. Same headline rate, much bigger invoice, and the trend is not an accident.
        speakable:
          "@type": SpeakableSpecification
          cssSelector:
            - .post-tldr
            - h1
        image: https://lukeocodes.dev/og.png?h=Why+Claude%27s+Token+Cost+Is+Higher+Than+Ever%2C+by+Design&d=The+Claude+token+cost+per+task+keeps+climbing+while+the+rate+card+barely+moves.+Default+thinking%2C+tokenizer+changes+and+quiet+cache+cuts+explain+the+bill.&e=Writing
        wordCount: 542
        timeRequired: PT3M
        publisher:
          "@id": https://lukeocodes.dev/authors/luke#person
      - "@type": BreadcrumbList
        itemListElement:
          - "@type": ListItem
            position: 1
            name: Home
            item: https://lukeocodes.dev
          - "@type": ListItem
            position: 2
            name: Writing
            item: https://lukeocodes.dev/blog
          - "@type": ListItem
            position: 3
            name: Why Claude's Token Cost Is Higher Than Ever, by Design
            item: https://lukeocodes.dev/claude-token-cost
      - "@type": FAQPage
        "@id": https://lukeocodes.dev/claude-token-cost#faq
        mainEntity:
          - "@type": Question
            name: Didn't Anthropic make Opus more efficient?
            acceptedAnswer:
              "@type": Answer
              text: "They did, once. Opus 4.5 in November 2025 was marketed around producing the same work with roughly 76% fewer output tokens. The reversal came after: newer models run extended thinking by default, so the token count per task went back up even though the per-token price looks similar."
          - "@type": Question
            name: What happened to prompt caching?
            acceptedAnswer:
              "@type": Answer
              text: The default cache TTL appears to have been cut from one hour to five minutes around March 2026, with no changelog entry. One developer reconstructed it from roughly 119,000 API calls and measured about a 26% overpayment that month. Shorter TTL means more cache misses, and cache misses are billed as fresh input.
          - "@type": Question
            name: Is the 10x claim measured?
            acceptedAnswer:
              "@type": Answer
              text: "That figure is mine, from my own usage: Opus 4.5 did work for roughly a tenth of the tokens a newer default-thinking model spends on the same task. I'm not presenting it as a published benchmark. The public numbers only confirm the direction, not that exact multiple."
---

[Luke Oliff.](/)

- [Blog](/blog)
- [CV](/cv)
- [Uses](/uses)
- [User manual](/user-manual)
- [Living with ADHD](/living-and-working-with-adhd)
- [RSS](/rss.xml)

[Luke Oliff · Writing](/blog)

# Why Claude's Token Cost Is Higher Than Ever, by Design

Aug 2026 · [Opinion](/blog/category/opinion) · 3 min read · [Luke Oliff](/authors/luke)

**TL;DR:** Opus 4.5 was the efficient one. Everything since burns more tokens per task with thinking on by default, a costlier tokenizer, and a prompt cache TTL that got quietly cut. Same headline rate, much bigger invoice, and the trend is not an accident.

The Claude token cost per unit of real work is higher than it has ever been, and the uncomfortable part is that this looks deliberate. The rate card barely moved. The number of tokens it takes to get the same answer went up a lot.

I'll say the strong version first, and label it honestly as mine. In my own logs, Opus 4.5 finished tasks for roughly a tenth of the tokens a newer default-thinking model spends on the same job. That's my measurement, not a study. But the public evidence all points the same way, so I'm comfortable standing on it.

## The efficiency high point, then the reversal

Opus 4.5, back in November 2025, was the efficient one. Anthropic marketed it around producing the same quality with about 76% fewer output tokens. That was real, and it was good.

Then the direction flipped. The models after it run extended "thinking" on by default, and thinking is billed. Secondary analyses put newer models at meaningfully higher token use for equivalent tasks, one estimate around 30% more, and the framing that stuck with me was blunt: the same rate card can still produce a larger invoice. When the model decides how much to think and thinking is your cost, "same price per token" is a sleight of hand.

## The cut nobody announced

Here's the one that actually bothers me. Around March 2026, the default prompt-cache TTL appears to have quietly dropped from one hour to five minutes. No blog post. No API version bump. A developer reconstructed the change from roughly 119,000 Claude Code API calls and measured about a 26% overpayment that month against the prior baseline.

Caching is where cost optimisation lives. A cache read is billed at a fraction of fresh input, so a shorter TTL means more misses, and every miss bills like new tokens. Trim the TTL and you raise everyone's bill without touching a single published price. That's not a bug. That's a lever.

## It keeps going

The pattern has more entries than one post can hold. Developers reported that a tokenizer change on a newer Opus raised token consumption 32 to 47% for identical work. Third-party frameworks that routed around API pricing got blocked. And in September 2026 Anthropic announced a "25% increase" to weekly limits that, measured against the temporary boost it replaced, [worked out to roughly a 17% cut](https://www.theregister.com/2026/01/05/claude_devs_usage_limits/). A cut dressed as a gift.

Any one of these you could wave off. Stack them, and there's a clear direction: more tokens per task, more misses per session, more friction for anyone trying to spend less. All while the headline price stays quotable.

## Why I think it's structural

Anthropic is burning at a scale that only closes if revenue per user climbs. Thinking-on-by-default climbs it. A shorter cache TTL climbs it. A pricier tokenizer climbs it. None of these require a price hike you'd notice on the pricing page, which is exactly why they work.

I still use Claude. It's very good. But I've stopped assuming the token count is a neutral fact about the model, and started treating it as a number someone at the company has an incentive to grow. If you're paying per token, you should too.

[← Newer: uv Is the Python Packaging Tool That Finally Clicked](/uv-python-packaging)

[All posts](/blog) →

## Search this site

`GET https://lukeocodes.dev/search.json?q=<words>` returns matching posts as JSON, each with its page URL and its markdown URL. Add `&limit=<1-50>` for more than 10. How it works: [/blog/search](https://lukeocodes.dev/blog/search.md).

```sh
curl "https://lukeocodes.dev/search.json?q=developer+relations"
```

---

© 2026 Luke Oliff. Posts live at the root of this site.

- [GitHub](https://github.com/lukeocodes)
- [LinkedIn](https://www.linkedin.com/in/lukeocodes)
- [X](https://x.com/lukeocodes)
- [Bluesky](https://bsky.app/profile/lukeocodes.dev)
- [Email](mailto:luke@lukeoliff.com)
- [RSS](/rss.xml)

## Cookies

This site sets one analytics cookie, from Google Analytics, and only in a browser that accepts the cookie banner. A client that does not run scripts sets no cookies, so there is nothing to accept or decline here.

[Cookie policy](/cookies)
