Large Models
The large-model topic tracks progress in foundation models themselves: new releases, the gap between open and closed source, the gap between Chinese and American models, and the economics of training and inference. Regular guests include Hugging Face engineers and practitioners, with a systematic review every quarter. The core questions: where has the capability frontier moved, how fast is the cost curve falling, and who is falling behind. Judgment over news — everyone can watch the launch events; we care about which improvements are real and which are marketing.
11 edited conversations, newest first.
EP61
Lawyer workflows · Legal judgment · Model evaluation · Engineering leverage
EP54
Consumer signals · recommendation flywheels · conversational advertising · user trust
EP53
Email as context · enterprise memory · closed ecosystems · local models
EP52
Public work · eighteen months · a hundred people · the infrastructure behind a talent map
EP48
Account pools · Multipliers · American women aged 45 to 55 · Shadow fleets
EP47
Two Three Gorges Dams · A trillion in depreciation · Finding MLCC among three thousand names · The radio was wrapped electricity
EP45
Four times leverage · Situational Awareness · Faith versus risk control · What did he learn
EP43
RSI · Wish-driven programming · A corrupted Einstein · LongCat and Hunyuan 3 — the 2026 midterm on large models
EP
Partly defined hardware · Products defined by users · The boundaries of recording · Robots that do real work
EP
Models with a short shelf life · Ninefold customer spending · A one-person drug company · The next coding opportunity is still coding
EP
After reading 19,262 résumés: the cities, schools, previous employers, functions and languages of the Big Three frontier labs — and how many Chinese nationals actually work there.