Hyper-π-bench: Evaluating agents that build agentsToday weβre open-sourcing hyper-π-bench, a new long horizon agent evaluation that measures how well models can not only act as an agent, but construct one.September 8, 2026
π-knowledge: benchmarking agents on real-world knowledgeπ-Knowledge measures how well agents can work through messy, evolving knowledge bases to complete complex, multi-step tasks. While models are improving, they still struggle to reliably use this information in practice, leaving a large gap to real-world performance.May 13, 2026
πΒ³-Bench: Advancing agent benchmarking to knowledge and voiceπΒ³-Bench is here. We've expanded agent evaluation to two new frontiers: knowledge retrieval and voice.March 18, 2026
Improving voice performance with post-trainingPost-training helps our customers' voice agents achieve shorter, clearer, and more human-like conversations.November 12, 2025
π-Bench leaderboard: compare, explore, and understand agent performanceIntroducing the π-Bench leaderboard β a community-driven platform where researchers can submit, verify, and compare results while exploring model behavior through interactive tools.October 13, 2025
