Stop counting tokens, start counting what ships
MIT Technology Review ·
Summary: When a developer gives an AI agent a coding task, the agent does more than answer once. It plans, generates code, runs tests, and repeats the process to correct errors. A single task can use 10 to 50 times more tokens than a basic chat prompt. The more useful the agent becomes, t
Read the full story on MIT Technology ReviewExplore more: Today in AI · Research papers · AI tools · Learn AI
Related stories
- The Decoder · 2 Nov 2026One brutal Minecraft night scared 50 AI agents into inventing chicken handoff protocols, a Fertilizer Officer, and fishing shifts
In experiments by Y Combinator Labs and Olam Labs, OpenAI's GPT-6.1 Sol model showed two very different sides. When 50 agents were tasked with peaceful civilization-building in Minecraft, they froze up in absurd bureaucracy. In direct competition against Claude Opus 5.5, they ind
- Anadolu Ajansı · 2 Nov 2026Claude developer Anthropic's quarterly revenue surges 12-fold to nearly $17B ahead of IPO
- Unite.AI · 2 Nov 2026OpenAI Launches Grid Protectors Program for Compute Campuses
On November 2, 2026, OpenAI launched the Grid Protectors Program, a national effort to arrange for future compute campuses to reduce electricity demand when grids are under strain. Georgia is the starting point, with a planned Effingham County campus and an agreement providing up
- Polygon.com · 2 Nov 2026Anthropic Is Officially Stepping In to Stop Claude from Cloning Games
- X · 2 Nov 2026Claude Opus 5.5 Wins Praise for Agent Building and Video Magic
- NYT Technology · 2 Nov 2026Inside Anthropic’s Hunt for A.I. Computing Power
The company went on a deal spree with tech giants, former crypto miners and others to obtain one of the artificial intelligence boom’s scarcest resources.
Who reported “Stop counting tokens, start counting what ships”?
MIT Technology Review, on 2 Nov 2026. The full article is at the source link on this page.