Open models now beat last year's frontier — so what moat remains for frontier labs?
andimarafioti · x · 2026-11-10
andimarafioti poses an industry-level question: a year ago the frontier was Opus 4.1 and GPT-5, and today's open-weight models beat all of them. Past a certain capability point, being better matters less — which is why labs sell cheaper, weaker tiers like Sonnet and Terra. Extrapolating: when open models beat today's frontier in a year, what moat will frontier labs actually have left?
More from AGI Musings
- Labs are keeping their most capable cyber, math and bio models internal, warns lvwerra — TheZachMueller · 2026-11-10
- In the age of vertiginous AI ascent, the stiff-drink-first movie trope makes more sense — StewartalsopIII · 2026-11-10
- Outside the Western bubble, AI use is widespread and positive, even in official orgs — technollama · 2026-11-10
- 'An expert in every domain in your pocket': Reddit user calls out pure anti-AI resentment — chessboardtable · 2026-11-10
- The History of the History of Algorithms: Practitioner Histories and Teleology — fkasummer · 2026-11-10
- Rat hippocampus maps other rats on the same cells it uses for self — AnnaCiaunica · 2026-11-10