Authors Omar Darwiche Domingues, Keenan Samway, Daniel D'souza, Marzieh Fadaee, Tan Yi-Chern, Nouha Dziri
Publication date 12 October 2026
Tldr We develop a framework for scaling inference-time compute with teams of coding agents by solving a compute allocation problem: how to allocate additional computation among planning, implementation, and verification stages to reach the reward-cost Pareto frontier. Our results offer a guide to allocating agent budgets where they count most, given task, cost, and latency constraints.
This page was built for publication: Scaling Inference-Time Compute with Teams of Coding Agents
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7400282)
