Ready Cohorts: Bounding GPU Opportunity and Avoiding Host Round Trips in LLM-Agent Control
Paper • 2608.12123 • Published • 2
None defined yet.
When Does Combining Language Models Help? A Co-Failure Ceiling on Routing, Voting, and Mixture-of-Agents Across 67 Frontier Models
AURA: Action-Gated Memory for Robot Policies at Constant VRAM