Two days after Signal65 launched its new Pinnacle benchmark for agentic AI, a new model release reversed one of its most consistent early findings. Ryan Shrout, President of Signal65, and Mitch Lewis join JV Roig of Kamiwaza.ai to discuss why Claude Fable 5.1 posted frontier-level performance with a fabrication rate other top models haven't matched, and what Pinnacle's early results reveal about evaluating AI for real enterprise work.

JV Roig, AI Platform Engineer at Kamiwaza, joins the Ryan Shrout & Mitch Lewis to discuss why real-world enterprise tasks need new agentic AI benchmarks, sharing key insights from recent Signal65 and Kamiwaza research on the KAMI index.