METR introduces a new metric to calculate exactly when AI agents become more expensive than humans
METR just put a price tag on AI agent economics. And the early numbers are sobering.
Why it matters
As AI agents move from research to deployment, understanding their true cost-of-ownership relative to human labor is becoming critical for enterprise decision-making. METR's 'expenditure horizon' metric attempts to quantify the breakeven point—but early benchmarks suggest we're not there yet.
The key facts
9 to knowMETR introduces 'expenditure horizon' metric for AI agent cost-effectiveness
Benchmark tested on NanoGPT speedrun with underwhelming early results
Metric designed to calculate dollar-denominated breakeven between AI agents and human labor
Framework has identified blind spots in current measurement approach
Next-generation models could shift economic viability picture
Early NanoGPT speedrun results described as underwhelming
Metric identifies blind spots in current agent evaluation frameworks
Newer model generations could shift cost-performance dynamics
Addresses agent vs. human labor economics—key decision point for enterprise AI deployment
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: METR's new metric, the "expenditure horizon," puts a dollar figure on how cost-effective AI agents are at solving problems. Early results on the NanoGPT speedrun are underwhelming, the metric has blind spots, and the newest generation of models could change the picture.