GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning

Researchers introduced GEPA (Genetic-Pareto), a prompt optimizer that uses natural language reflection to improve LLM performance through trial and error. Across six tasks, GEPA outperformed reinforcement learning methods like GRPO by 6% on average and up to 20%, while using up to 35x fewer rollouts. The approach also surpassed the leading prompt optimizer MIPROv2 by over 10% and showed promise for code optimization.

Computer Science Computation and Language Submitted on 25 Jul 2025 v1 https://arxiv.org/abs/2507.19457v1 , last revised 14 Feb 2026 this version, v2 Title:GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning View PDF /pdf/2507.19457 Abstract:Large language models LLMs are increasingly adapted to downstream tasks via reinforcement learning RL methods like Group Relative Policy Optimization GRPO , which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA Genetic-Pareto , a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories e.g., reasoning, tool calls, and tool outputs and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6% on average and by up to 20%, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10% e.g., +12% accuracy on AIME-2025 , and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at this https URL . Submission history From: Lakshya A Agrawal view email /show-email/af5f91b3/2507.19457 Fri, 25 Jul 2025 17:42:32 UTC 1,632 KB v1 /abs/2507.19457v1 v2 Sat, 14 Feb 2026 11:42:30 UTC 1,650 KB Current browse context: cs.CL References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .