gepa

back during a reflection step and proposes targeted fixes, not blind mutations. Candidates are selected via a Pareto frontier across metrics/examples, so a candidate that's best at one thing survives even if its average is mediocre. Two ideas distinguish this from AlphaEvolve/OpenEvolve/ShinkaEvolve-style LLM evolution: (1) diagnostic feedback is a first-class API concept rather than a framework-specific mechanism, and (2) the API unifies three optimization modes — single-task search (solve one hard problem), multi-task search (solve related problems with cross-transfer), and generalization (build artifacts that transfer to unseen inputs). Prior frameworks only express mode 1. We tested across 8 domains. Selected results: Coding agent skills: Learned repo-specific skills push Claude Code to near-perfect task completion and make it 47% faster Cloud scheduling: Discovered algorithms that cut costs 40%, topping the ADRS leaderboard over expert heuristics and other LLM-evolution frameworks Agent architecture: Evolved a 10-line stub into a 300+ line ARC-AGI agent, improving Gemini Flash from 32.5% → 89.5% Circle packing (n=26): Outperforms AlphaEvolve's published solution Blackbox optimization: Generated problem-specific solvers matching or exceeding Optuna across 56 EvalSet problems CUDA kernels: 87% match or beat baseline; multi-task mode outperforms dedicated single-task runs ``` pip install gepa ``` Blog with full results and runnable code for all 8 case studies: https://gepa-ai.github.io/gepa/blog/2026/02/18/introducing-o... GitHub: https://github.com/gepa-ai/gepa

reads this feedback during a reflection step and proposes targeted fixes, not blind mutations. Candidates are selected via a Pareto frontier across metrics/examples, so a candidate that's best at one thing survives even if its average is mediocre. Two ideas distinguish this from AlphaEvolve/OpenEvolve/ShinkaEvolve-style LLM evolution: (1) diagnostic feedback is a first-class API concept rather than a framework-specific mechanism, and (2) the API unifies three optimization modes — single-task search (solve one hard problem), multi-task search (solve related problems with cross-transfer), and generalization (build artifacts that transfer to unseen inputs). Prior frameworks only express mode 1. We tested across 8 domains. Selected results: Coding agent skills: Learned repo-specific skills push Claude Code to near-perfect task completion and make it 47% faster Cloud scheduling: Discovered algorithms that cut costs 40%, topping the ADRS leaderboard over expert heuristics and other LLM-evolution frameworks Agent architecture: Evolved a 10-line stub into a 300+ line ARC-AGI agent, improving Gemini Flash from 32.5% → 89.5% Circle packing (n=26): Outperforms AlphaEvolve's published solution Blackbox optimization: Generated problem-specific solvers matching or exceeding Optuna across 56 EvalSet problems CUDA kernels: 87% match or beat baseline; multi-task mode outperforms dedicated single-task runs ``` pip install gepa ``` Blog with full results and runnable code for all 8 case studies: https://gepa-ai.github.io/gepa/blog/2026/02/18/introducing-o... GitHub: https://github.com/gepa-ai/gepa

hink about these things and come up with optimal prompt and corrections to it is a very hard problem to solve. [1] https://github.com/humanlayer/humanlayer/blob/main/.claude/c... [2] https://litestream.io/guides/vfs/#when-to-use-the-vfs [3] https://docs.boundaryml.com/guide/baml-advanced/prompt-optim... [4]https://github.com/gepa-ai/gepa