Frameworks, sandboxes and the evidence on when many agents beat one.
Pantone 17-5641 Emerald · Color of the Year 2013 · 3 pieces
AI agent frameworks get ranked three ways, by downloads, by GitHub stars and by production deployments, and the three rankings disagree; here is what the June 2026 data says about LangGraph, CrewAI, the OpenAI Agents SDK and the platforms closing in on them.
6 min · 13 sources
An August 2026 benchmark of AI agent sandboxes put cold starts between 0.27 and 5.06 seconds and the cost of 1,000 ten-minute agent loops between $11 and $53, numbers that decide which runtime an agent can afford and which isolation model it must accept.
8 min · 14 sources
A Google and MIT study of multi-agent systems measured gains of 81% on parallel financial analysis and losses of up to 70% on sequential planning, which makes task topology and the coordination bill, rather more than agent count, the variables that decide whether a swarm beats a soloist.
7 min · 14 sources