AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms - Google DeepMind

New AI agent evolves algorithms for math and practical applications in computing by combining the creativity of large language models with automated evaluators

AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms — Google DeepMind Skip to main content May 14, 2025 Science AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms AlphaEvolve team Share New AI agent evolves algorithms for math and practical applications in computing by combining the creativity of large language models with automated evaluators Large language models (LLMs) are remarkably versatile. They can summarize documents, generate code or even brainstorm new ideas. And now we’ve expanded these capabilities to target fundamental and highly

saved by

• Ratan Kaliani
• Chris Wong
• Information Flow
• Sarah Pan
• Lydia Nottingham

related reading

• AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms — Google DeepMinddeepmind.google
• [2506.13131] AlphaEvolve: A coding agent for scientific and algorithmic discoveryarxiv.org
• AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fieldsdeepmind.google
• Explore | alphaXivalphaxiv.org
• Competitive programming with AlphaCode — Google DeepMinddeepmind.com
• CodeEvolve: an open source evolutionary coding agent for algorithmic discovery and optimizationalphaxiv.org
• [2203.07814] Competition-Level Code Generation with AlphaCodearxiv.org
• [2507.19457] GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learningarxiv.org
• Explore | alphaXivalphaxiv.org
• The Unreasonable Effectiveness of LLMs in Mathematicschrishayduk.com
• Evo 2 Can Design Entire Genomesasimov.press
• Lluminatejoelsimon.net