Google Research· ResearchResearch paper

Towards a science of scaling agent systems: When and why agent systems work

Google Research publishes a paper introducing the first quantitative scaling principles for AI agent systems. The study evaluates 180 agent configurations across five architectures and finds that multi-agent coordination improves performance on parallelizable tasks but degrades it on sequential ones.

From the source

In our new paper, “Towards a Science of Scaling Agent Systems”, we challenge this assumption. Through a large-scale controlled evaluation of 180 agent configurations, we derive the first quantitative scaling principles for agent systems, revealing that the "more agents" approach often hits a ceiling, and can even degrade performance if not aligned with the specific properties of the task.

research.google

Announced

28 Jan 2026 11:00 UTC