Stop counting tokens start counting what ships
Original title: Stop counting tokens, start counting what ships
AI coding agents should be measured by cost per merged pull request rather than tokens consumed to track value delivered.
AI coding agents consume significantly more tokens than basic prompts leading to high costs for large tasks. A more effective metric is cost per merged pull request which measures useful work shipped. Hybrid approaches using open-weight models for routine tasks can reduce costs by 30-50 percent.
- Gartner predicts AI coding token costs will surpass developer salaries by 2028
- Cost per merged pull request is recommended metric
- HPE Private Cloud AI reduced token costs 30 to 50 percent
- Top 5 percent users generate most consumption
- Products
- HPE Private Cloud AIClaude Code
- Organizations
- Gartner
- AI models
- Claudeopen-weight models
From the original article
4 min read at technologyreview.com
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.
By Affan Syed · technologyreview.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.