Today for AI
HOT RADAR
agentHEAT 14.9°

AI CURATED HOT STORY · 11/12/2026

CMU and JetBrains Study Finds Token Spend Poorly Predicts AI Agent Output

1 reports archived1 independent sourcesupdated 11/12/2026, 01:13:42
Synthesis & Latest Updates
1 Sources Cross-Validated

A joint Carnegie Mellon and JetBrains study finds that engineers' token spend is only weakly linked to shipped output: for the same engineer, doubling spend corresponded to roughly 10% more commits. The work linked five weeks of coding-agent token usage to commits from 404 JetBrains engineers, backed by a 412-person company survey. The authors recommend using token spend only for budgeting and judging agent-assisted work by task and outcome evidence.

LATEST/A joint Carnegie Mellon and JetBrains study finds that engineers' token spend is only weakly linked to shipped output: for the same engineer, doubling spend corresponded to roughly 10% more commits. The work linked five weeks of coding-agent token usage to commits from 404 JetBrains engineers, backed by a 412-person company survey. The authors recommend using token spend only for budgeting and judging agent-assisted work by task and outcome evidence.

Hourly heat curve

6 fully observed hours
Now
14.9
Peak
16.902:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·78 pts

    CMU and JetBrains Study Finds Token Spend Poorly Predicts AI Agent Output

    • Doubling token spend for the same engineer yielded only about 10% more commits
    • Among weeks with similar commit counts, token usage varied by 39 to 240 times
    • Technical staff reported using agents beyond coding in nearly 9 of 10 evaluations
    • Token spend should inform budgeting only; output should be judged by task and outcome evidence