HaiAI123

Curated Global AI Tools Directory

NewIndustry update
New

Longer Claude Code sessions cost more as tokens pile up

10/27/2026 — 10/27, 19:20·1 sources (all)·1 reports (all)

Key takeaways

  • A developer analyzed a 236-request Claude Code session on Opus 5.5, showing that every message resends the full conversation, so requests in…

Background & analysis

On October 27, 2026, a developer writing on DEV Community broke down one of their own Claude Code sessions and argued that longer sessions get more expensive over time, and that caching does not stop the cost from accumulating.

According to the post, every message in a Claude Code session sends the whole conversation back to the model. Prompt caching makes that re-read cheap per token, the author explains, but cheap is not free, and the longer the session runs, the more there is to re-read.

The figures cited come from the author's own session: 236 requests in total, running on Opus 5.5. By the last quarter of that session, each request was carrying about 3.9 times the tokens of a request in the first quarter. The author also put together a short Python script so readers can see the same growth curve in their own sessions.

At this point only this single report is available, and there is no response from Anthropic or any further development.

AI-generated from 1 reports · updated 4 hours ago

Related AI tools

24-hour heatHeat index 89 · peak 100 · 4h ago
24 hours agonow

Reports on this story headlines open the original

Today
  1. A developer broke down one of his own Claude Code sessions (236 requests, Opus 5.5) and found that by the final quarter, each request carried about 3.9 times the tokens of an early request, because every message resends the whole conversation. Prompt caching makes each re-read cheaper per token, but not free. He published a short Python script so others can plot the same curve on their own sessions.

    DEV Community · AIAI score 53

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →