GEPA: reflective prompt evolution that beats RL on few rollouts
ICLR 2026 oral: Genetic-Pareto prompt optimization that diagnoses trajectories in natural language—strong results including code-optimization search.
Source: GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
GEPA argues that language is a denser learning medium than sparse scalar rewards. It samples trajectories (reasoning, tool calls, outputs), reflects in natural language, proposes prompt updates, and keeps complementary lessons from a Pareto frontier of past attempts.
Across six tasks it beats GRPO by ~6% on average (up to 20%) with up to 35× fewer rollouts, and beats MIPROv2 by over 10%. They also show it as an inference-time search strategy for code optimization.
Why it matters for code review
Prompt-and-scaffold tuning for review agents usually burns tokens on blind A/B. GEPA’s pattern—diagnose the trajectory, keep the complementary lessons—maps cleanly onto review harnesses where you want better reviewer instructions without RL infrastructure. The Pareto keep-set is especially relevant when different review failure modes need different prompt patches.