AlphaEvolve improves algorithms — and Google's own AI stack

Wednesday 14 May 2025 · significance ●●●●○

A Gemini-powered coding agent discovers a faster 4×4 complex matrix multiplication and recovers ~0.7% of Google's worldwide compute.

What happened and why it matters

AlphaEvolve improved on Strassen's 1969 result for 4×4 complex matrices, matched or beat the state of the art on ~20% of 50+ open maths problems, sped up a key Gemini training kernel by 23% (cutting overall training time ~1%) and optimised data-centre scheduling. It is one of the clearest concrete examples of AI improving the systems that train AI.

Sources

AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms Google DeepMind · 14 May 2025

Effect on the Gapnium Coefficient

+0.0078

0.2311 → 0.2388

Adjustments

gapnium.recursion +0.025 AI directly sped up training of its own successor models and freed fleet compute.
gapnium.discovery +0.015 Improved a 56-year-old algorithmic record and several open problems.
gapnium.compute +0.005 Efficiency recovered across a hyperscale fleet.

Review

Last reviewed by AI agent: 7 Nov 2026
Model: seed:claude-opus-5-5
Agent confidence: 0.75
Human reviewed: —

Initial historical seed. Scores assigned by the methodology rubric; awaiting first scheduled re-review.

Feedback on this scoring

Hover the chart to read values · click a marker to open its articles