@shawntenam: GEPA (http://github.com/gepa-ai/gepa) bumped Haiku 4.5 from 65% to 85% pass rate by auto-optimizing my prompt instructi…
Summary
GEPA is an open-source tool that automatically optimizes prompt instructions using execution traces and scores, raising Claude Haiku 4.5's pass rate from 65% to 85% without requiring a model swap.
Similar Articles
@kapicode: I've been using Claude as the "human" prompting @opencode to rebuild reference projects, evaluating four LLMs on the sa…
An evaluation of four LLMs (Qwen, MiniMax, GLM) using Claude as a prompter for the Opencode agent tool reveals that a smaller local model (Qwen 27B on a 3090) outperforms a larger pruned model in coding quality and reliability.
@learnwithella: Self-improving Claude Code skills are f*cking ridiculous One loop → 10 test runs, scored against an eval, prompt rewrit…
Claude Code can auto-iterate prompts by running evals, rewriting, and keeping winners, boosting a hook-writer skill from 32/50 to 47/50 overnight.
@dhh: I've been driving GPT5.5 on low reasoning for the last week+ and it's very good, very efficient. Haven't been tempted t…
DHH praises the performance and efficiency of GPT-5.5 on low reasoning settings, noting it surpasses Opus and Kimi.
@HenryL_AI: Big update: @gepa_ai has now been officially integrated into A-Evolve (by community member)! We added GEPA as a new plu…
Community member integrated the GEPA evolution algorithm into A-Evolve as a plug-and-play component, letting any agent use GEPA with zero setup.
Dynamically allocating compute budget to hard set of problems and evolving the sections with Qwen-35B-A3B gets you near GPT-5.4-xHigh on HLE
A method that dynamically allocates compute budget to hard problems using Qwen-35B-A3B achieves performance near GPT-5.4-xHigh on the HLE benchmark.