EnvOpt Automatically Cuts Agent Environment Costs by 48%
Start here
Researchers at the University of Chicago, UC Berkeley Sky Computing Lab and Bespoke Labs have published EnvOpt, a method that automatically restructures an agent's stationary environment to reduce cost. Across five benchmarks, EnvOpt cut target-agent cost on unseen test sets by an average of 48% while pass rates stayed the same or improved. On AutomationBench it beat Meta-Harness by 9.67% in pass rate at only 15% of the cost, and the optimized environments transferred to other harnesses and models, including a 75.7% average cost reduction on AutomationBench. Most of the savings come from input tokens, mainly tool execution results. An offline optimization run on AutomationBench costs about USD $150, and the authors report it saves 77.5% of the cost of running Claude Opus 4.8.
Generated from reports · updated 33 min ago
Report timeline
Follow the reports to see the event from different sides.
- Hacker News · AIEnvOpt: Automated Optimization for Agent Environments
EnvOpt, a method from University of Chicago, UC Berkeley Sky Computing Lab and Bespoke Labs, automatically restructures an agent's stationary environment to cut cost, reducing target-agent cost on unseen test sets by an average of 48% across five benchmarks with higher or similar pass rates. On AutomationBench it beats Meta-Harness by 9.67% pass rate at only 15% of its cost, and the optimized environments transfer to other harnesses and models, including a 75.7% average cost reduction on AutomationBench. Most savings come from input tokens, mainly tool execution results, and an offline optimization on AutomationBench costs around USD $150 while saving 77.5% of the cost of running Claude Opus 4.8.
Heat for this event
Not enough continuous observations yet to draw a trend.