Huawei EvoTrias Decouples Search Planning from Coding to Cut Token Cost 64%
Huawei proposes EvoTrias, a code-evolution agent framework that assigns search planning, code writing, and scoring to three separate owners: a controller, a coding agent, and an evaluator the agent cannot read. It led 6 of 7 math tasks and cut token cost per run by 64% versus the 4-agent CORAL, while lifting solver quality 13.2% on 300 instances of the 680K-line C++ OR-Tools repository, against -0.1% for one-shot runs of the same agent. The takeaway: hide the scorer from the agent and keep search planning out of the coding agent.
Full Report
Rohan Paul (@rohanpaul_ai) · 2026-11-16 19
New Huawei paper finds that code-evolution agents do better and spend less when search planning, code writing, and scoring are handled by 3 separate owners.
AlphaEvolve-style systems improve programs by writing code, scoring it, and picking what to try next. Some give the code writer little freedom, others let 1 coding agent steer the search at high token cost, and none hides the scorer.
EvoTrias gives each job its own owner. A controller steers the search, a coding agent like Claude Code writes code, and an evaluator the agent cannot read scores it.
On 7 math tasks it led 6 and cut token cost per run by 64% versus 4-agent CORAL. On OR-Tools, a 680K-line C++ repository, it lifted solver quality 13.2% on 300 instances, while 1-shot runs of the same agent averaged -0.1%.
For code-optimizing agent loops, hide the scorer from the agent and keep search planning out of the coding agent.
