AlphaEvolve

<h2 id="design">Design</h2>
<p>AlphaEvolve aims to autonomously discover and refine algorithms through a combination of <a href="/facts/Large_language_models/WnogWVJY">large language models</a> (LLMs) and evolutionary computation.<a class="footnote-ref" id="fnref:3" href="#fn:3"><sup>3</sup></a>
</p><p>AlphaEvolve needs an evaluation function with metrics to optimize, and an initial algorithm. At each step, AlphaEvolve uses the LLM to produce variants of the existing algorithms, and then selects the most effective ones.<a class="footnote-ref" id="fnref:4" href="#fn:4"><sup>4</sup></a>
</p><p>Unlike domain-specific predecessors like <a href="/facts/AlphaFold/DwBhppKC">AlphaFold</a> or <a href="/facts/AlphaTensor/VrYWxfDt">AlphaTensor</a>, AlphaEvolve is designed as a general-purpose system. It can operate across a wide array of scientific and engineering tasks by automatically modifying code and optimizing for multiple objectives. Its architecture allows it to evaluate code programmatically, reducing reliance on human input and mitigating risks such as hallucinations common in standard LLM outputs.<a class="footnote-ref" id="fnref:5" href="#fn:5"><sup>5</sup></a>
</p>
<h2 id="achievements">Achievements</h2>
<p>According to Google, across a selection of 50 open <a href="/facts/Mathematical_problems/lFMR51de">mathematical problems</a>, the model was able to rediscover state-of-the-art solutions 75% of the time and discovered improved solutions 20% of the time, for example advancing the <a href="/facts/Kissing_number_problem/UE3mtV0d">kissing number problem</a>.<a class="footnote-ref" id="fnref:6" href="#fn:6"><sup>6</sup></a>
</p><p>AlphaEvolve was also used to optimize Google's computing ecosystem. Improved data center scheduling heuristics, enabled the recovery of 0.7% of stranded resources. It was also used to optimize <a href="/facts/Tensor_Processing_Unit/aVlg0LFF">TPU</a> circuit design and <a href="/facts/Gemini_(language_model)/eD5D8xBo">Gemini</a>'s training <a href="/facts/Matrix_multiplication_algorithm/VrYWxfDt">matrix multiplication</a> kernel.<a class="footnote-ref" id="fnref:7" href="#fn:7"><sup>7</sup></a>
</p>
<h2 id="see-also">See also</h2>
<ul><li><a href="/facts/Gemini_(chatbot)/Kwfa4rMW">Gemini (chatbot)</a></li>
<li><a href="/facts/Strassen_algorithm/rrIGRSNp">Strassen algorithm</a></li></ul>

<h2 id="external-links">External links</h2>
<ul><li><a href="https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/AlphaEvolve.pdf">AlphaEvolve white paper</a></li></ul>

<h2 id="references">References</h2>

<ol>
<li id="fn:1"><p>"AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms". Google DeepMind. 2025-05-14. Retrieved 2025-05-14. <a href="https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/" target="_blank">https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/</a> <a href="#fnref:1" class="footnote-back-ref">↩</a></p></li>
<li id="fn:2"><p>Whitwam, Ryan (2025-05-14). "Google DeepMind creates super-advanced AI that can invent new algorithms". Ars Technica. Retrieved 2025-05-17. <a href="https://arstechnica.com/ai/2025/05/google-deepmind-creates-super-advanced-ai-that-can-invent-new-algorithms/" target="_blank">https://arstechnica.com/ai/2025/05/google-deepmind-creates-super-advanced-ai-that-can-invent-new-algorithms/</a> <a href="#fnref:2" class="footnote-back-ref">↩</a></p></li>
<li id="fn:3"><p>Nuñez, Michael (2025-05-14). "Meet AlphaEvolve, the Google AI that writes its own code—and just saved millions in computing costs". VentureBeat. Retrieved 2025-05-14. <a href="https://venturebeat.com/ai/meet-alphaevolve-the-google-ai-that-writes-its-own-code-and-just-saved-millions-in-computing-costs/" target="_blank">https://venturebeat.com/ai/meet-alphaevolve-the-google-ai-that-writes-its-own-code-and-just-saved-millions-in-computing-costs/</a> <a href="#fnref:3" class="footnote-back-ref">↩</a></p></li>
<li id="fn:4"><p>Tardif, Antoine (2025-05-17). "AlphaEvolve: Google DeepMind's Groundbreaking Step Toward AGI". Unite.AI. Retrieved 2025-05-25. <a href="https://www.unite.ai/alphaevolve-google-deepminds-groundbreaking-step-toward-agi/" target="_blank">https://www.unite.ai/alphaevolve-google-deepminds-groundbreaking-step-toward-agi/</a> <a href="#fnref:4" class="footnote-back-ref">↩</a></p></li>
<li id="fn:5"><p>Whitwam, Ryan (2025-05-14). "Google DeepMind creates super-advanced AI that can invent new algorithms". Ars Technica. Retrieved 2025-05-17. <a href="https://arstechnica.com/ai/2025/05/google-deepmind-creates-super-advanced-ai-that-can-invent-new-algorithms/" target="_blank">https://arstechnica.com/ai/2025/05/google-deepmind-creates-super-advanced-ai-that-can-invent-new-algorithms/</a> <a href="#fnref:5" class="footnote-back-ref">↩</a></p></li>
<li id="fn:6"><p>Nuñez, Michael (2025-05-14). "Meet AlphaEvolve, the Google AI that writes its own code—and just saved millions in computing costs". VentureBeat. Retrieved 2025-05-14. <a href="https://venturebeat.com/ai/meet-alphaevolve-the-google-ai-that-writes-its-own-code-and-just-saved-millions-in-computing-costs/" target="_blank">https://venturebeat.com/ai/meet-alphaevolve-the-google-ai-that-writes-its-own-code-and-just-saved-millions-in-computing-costs/</a> <a href="#fnref:6" class="footnote-back-ref">↩</a></p></li>
<li id="fn:7"><p>Nuñez, Michael (2025-05-14). "Meet AlphaEvolve, the Google AI that writes its own code—and just saved millions in computing costs". VentureBeat. Retrieved 2025-05-14. <a href="https://venturebeat.com/ai/meet-alphaevolve-the-google-ai-that-writes-its-own-code-and-just-saved-millions-in-computing-costs/" target="_blank">https://venturebeat.com/ai/meet-alphaevolve-the-google-ai-that-writes-its-own-code-and-just-saved-millions-in-computing-costs/</a> <a href="#fnref:7" class="footnote-back-ref">↩</a></p></li>
</ol>

AlphaEvolve open-in-new

AlphaEvolve