[arXiv]score: 0.28MoMHa Optimizes LLM Harnesses for Accuracy Safety and CostSeptember 28, 2026MoMHa uses a single-phase joint-reward proposer to optimize the Python code surrounding LLMs, balancing accuracy, behavioral safety, and token consumption. Testing across 17 domains shows that joint-reward optimization outperforms sequential accuracy-then-token optimization strategies.HOW THIS AFFECTS YOU●builderYou can optimize your production wrappers to reduce token costs without sacrificing accuracy or safety.●researcherThis introduces a new multi-objective search framework for automating agentic harness design.read original ↗arxiv.orgDAILY DIGEST_all newsbuilderresearcherfounderinvestordesignerpolicyhealthsubscribe →you don't check 9 sources — we do. one email every morning, read in 2 min. free. unsubscribe anytime. privacy← back to feed