1 articles with this tag
Frontier models score 25% on Polymath's Horizon-SWE benchmark. That gap, between what today's best agents can do and what software teams actually need, is the market Polymath is building for.
