Prompt EngineeringLakshya A Agrawal et al. (UC Berkeley, Stanford and others)UC BerkeleyStanford University

GEPA published

GEPA published — mutates prompts through natural-language reflection and evolves them with Pareto-based candidate selection

Why it matters

Argues that prompt evolution can compete with RL on sample efficiency

Source — Primary sources first; secondary reports are labelled “Press”

Paperarxiv.orghttps://arxiv.org/abs/2507.19457

Added in Week 35, 2026 · See that week

Same week

Voyage AI releases voyage-context-3OfficialSK Telecom open-sources A.X 3.1OfficialAlibaba releases Qwen3-CoderOfficialGemini Deep Think reaches IMO 2025 gold-medal standard (5 of 6 problems, 35/42)Official

Nearby on the same axis

Is Chain-of-Thought Reasoning of LLMs a Mirage?PaperChain of Thought Monitorability position paper publishedPaperChain-of-Thought Is Not Explainability publishedOfficialA. Lawsen publishes Comment on The Illusion of ThinkingPaper