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

*Source* · `src-deepmind-2025-alphaevolve-blog` · status: **n/a**

| field | value |
|---|---|
| kind | Technical report / blog |
| authors | Google DeepMind |
| venue | Google DeepMind blog |
| date | 2025-05-14 |
| url | https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/ |
| accessed | 2026-10-24 |

## Notes

Borg scheduling heuristic recovers about 0.7% of Google fleet compute; 4x4 complex matrix multiply in 48 multiplications; over 50 open problems: ~75% matched, ~20% improved; Gemini kernel 23% faster.

## Backlinks (computed)

**cited by:** [LLM-guided evolutionary program search](../technique/alphaevolve-algorithm-discovery.md)

## Citation

Google DeepMind (2025-05-14). *AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms*. Google DeepMind blog https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/

---
Provenance: created in run `2026-10-24-techniques-sim-math`.