YC Paper Club's AI inference session planned for October 21, 2026
A post promoting the session says well-tuned inference could differ from naive setups by 100-fold in latency and cost.
TLDR
An October 8 post announced a YC Paper Club session on AI inference for October 21, 2026. It argues that inference is now as complex and important as training, and that tuning could make a 100-fold difference in latency and cost compared with naive inference. The speaker list names Dan Fu of Together AI, Jeffrey Morgan of Ollama and Steven Arellano of Wafer, with a SGLang/RadixArk speaker still to be named.
Combined views
2.7K
1 Source, first seen ago
