Urgent.News

What's breaking now, across thousands of outlets.

AI

Claude Code now shows exactly why it burned your tokens, and I finally stopped guessing

The /cost is quite helpful.

Claude Code now shows exactly why it burned your tokens, and I finally stopped guessing

Claude Code now provides a cost command that shows exactly how many tokens have been used. A user discovered that leaving conversations open and returning to them later was costing them a significant number of tokens.

The new cost command provides detailed information on token usage, allowing users to track their costs. According to Dev.to, a plugin called quota-audit can also be used to break down usage per project and per skill, showing the exact times a rate limit was hit and which project was blocked.

Some companies, including Meta and Microsoft, are working to reduce their employees' use of Claude. Meta employees using Claude Code have dropped to around 30,000 from 60,000 earlier in the year, according to The Information, as reported by Techmeme.

Brief written by urgent.news from XDA Developers, Techmeme, AWS Machine Learning, Dev.to — 4 reports on this story. Machine-written — may contain errors; check the original before relying on it.

Read the original at xda-developers.com →

More in AI

All the drama around AI’s takeover of mathematics

This past year, OpenAI, Anthropic, and other labs have announced breakthroughs on numerous long-standing mathematical problems, in some cases pushing well beyond what researchers expected current systems to be capable of — including resolving one of the famous Millennium Prize problems.

Palestinian invests in AI in São Paulo

Aws Alnabulsi was born in Jordan and moved to Palestine at age 4. Today, his company, Coretava, has an office in São Paulo and employees in the Arab country.

  • Palestinian-American engineer invests in AI platform TavyoOS
  • Coretava seeks $5 million for technology and growth
  • 11 Palestinian professionals work at São Paulo office

The 0-Click AI Attack: How Indirect Prompt Injection Hijacks AI Agents

"How attackers can compromise AI agents without ever touching the AI interface—by hiding instructions inside documents, emails, web pages, RAG content, and tool responses." The Trust Propagation Layer…

  • Indirect Prompt Injection attacks AI agents without direct interface interaction.
  • Malicious instructions can hide in emails, documents, web pages, or tool responses.
  • Trust metadata should track data origin to prevent untrusted data influence.

More from Monday 5 October →