A Jev that can ask Opus for help gives Opus accuracy at 4x lower cost(hightouch.com)
568 points by theanomaly | 142 comments
Hightouch blog post benchmarking cheap decision models (Jev, GPT-6 Luna) against frontier LLMs for grading identity resolution, finding a Jev+Opus hybrid matches Opus accuracy at 4x lower cost.
