TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient
TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient
TokenEconomizer is an open-source plugin from UC Berkeley, Microsoft Research, and Microsoft Security that learns from an agent's past sessions in a repository to cut waste without modifying the agent or its model. Across 277 SWE-Bench tasks and five agent-model combinations, it reduced average cost by 18.4–30.9% in all 30 repository-combination pairs, with task success rates broadly stable. It works through micro-skills, tool pruning, and failure-loop intervention, and also lowered costs 16.0–21.2% on Web-Bench and 6.4–19.6% on security auditing.
It offers a concrete, evaluated method for cutting AI coding agent token costs without changing the underlying model, with per-repository results and an open-source release.
Source: Microsoft Research · microsoft.com