Solve harder problems with AlphaEvolve, now available to everyone on Google Cloud

July 9, 2026

What’s new: Today, AlphaEvolve is generally available (GA) on Gemini Enterprise Agent Platform. AlphaEvolve is a code optimization and discovery agent built on top of Gemini that helps solve the hardest algorithmic problems for your business and research. It has been tested in diverse domains like logistics, semiconductors, genomics, high performance computing, and financial services during our early access program.

Deploying AlphaEvolve within your environment follows a structured four-step process:

1. Define: Provide a baseline seed algorithm and problem definition, together with background knowledge.
2. Measure: Establish a scoring function to objectively score candidate programs.
3. Optimize: Use AlphaEvolve’s agentic harness to generate optimized code.
4. Apply: Deploy the resulting, highly optimized algorithm directly into production workloads.

How organizations are using AlphaEvolve

Customer results cited include:

• BASF: Digital twin for global supply chains, improving planning and forecasting models by over 80%.
• Coolblue: 5%+ WMAPE improvement on 28-day demand forecasting in ~200 iterations.
• FM Logistic: 10.4% warehouse routing improvement, saving 15,000+ km in staff travel.
• Infineon: Chip design lifecycle optimization including surrogate modelling.
• JetBrains: 15-20% IDE performance improvement.
• Kinaxis: 22%+ forecasting accuracy gains, 90%+ runtime reduction on benchmarks.
• Klarna: Doubled ML training pipeline throughput while improving model quality under regulated constraints; explored ~6,000 candidate programs over three weeks.
• Oak Ridge National Laboratory: AlphaEvolve on Frontier exascale supercomputer for mixed-precision GPU kernel optimization.
• PacBio: 30% reduction in variant detection errors for DeepConsensus.
• WPP: 5-10% prediction accuracy lift on campaign performance models.

Hardening Google’s own infrastructure

Google has integrated AlphaEvolve to optimize next-generation TPU silicon design, reduce Google Spanner write amplification by 20%, cut software storage footprints by nearly 9% via compiler strategies, boost natural disaster risk prediction accuracy by 5% across 20 categories, and discover quantum circuits with 10x lower error rates on Willow processor.

Pushmeet Kohli, Chief Scientist, Google Cloud & VP Science at Google DeepMind: “AI is moving beyond acting as a productivity assistant that accelerates how we work to a discovery engine that expands what we can achieve.”

Getting started

Requires a seed program (baseline algorithm with optimizable segments) and a client-side evaluator that scores mutated candidates. AlphaEvolve Skill available for IDEs including Antigravity and Claude Code.