On May 14, 2025, DeepMind announced AlphaEvolve, a coding agent that uses Gemini models inside an evolutionary loop to discover and improve algorithms. It is the successor to FunSearch: where FunSearch evolved short functions, AlphaEvolve evolves whole programs, using Gemini Flash for breadth and Gemini Pro for depth, with automated evaluators verifying every candidate.
Its headline mathematical result was a way to multiply two 4x4 complex-valued matrices using 48 scalar multiplications, the first improvement over Strassen’s 1969 algorithm for that case in 56 years. Applied to a set of more than 50 open problems across mathematical analysis, geometry, combinatorics, and number theory, AlphaEvolve rediscovered the best known solution in about 75 percent of cases and improved on it in roughly 20 percent, including a new lower bound for the kissing number problem in 11 dimensions.
AlphaEvolve also delivered practical gains inside Google. A scheduling heuristic it found has run in production for over a year and on average continuously recovers 0.7 percent of Google’s worldwide compute resources, and it contributed optimizations to chip design and AI training.
AlphaEvolve shows the FunSearch pattern - a creative model bounded by a strict evaluator - scaling up from isolated puzzles to real engineering and open research problems.