Fine-tuned open models cost 95% less and beat frontier APIs — are AI labs' moats gone?
ayushtweetshere · x · 2026-09-14
A viral thread argues custom fine-tuned open-source models trained on proprietary data cost 95% less than frontier models, perform better, and train in under 48 hours — leaving frontier labs without a moat even at the enterprise level. The author accuses labs of seeding FUD (e.g., "hacking agent swarm" demos) as their only defense, with repliers comparing it to the Phoebus lightbulb cartel: regulation itself becomes the moat, and what Dario and Sam reportedly fear. Numbers are unverified but the open-source-vs-API business debate is resonating.
Related event: Open-Source Fine-Tuning at 5% Cost Sparks Moat Debate for Closed Labs(11 posts)→
More from AGI Musings
- DeepMind's Economics of Intelligence: AI reshapes economy via agent coordination, not solo model smarts — Scobleizer · 2026-10-31
- Claude can build pro software for ~$20k in tokens — and TAM shrinks to match — bigblueboo · 2026-10-31
- Will AI kill the 9-5? A long read traces the workday back to Josiah Wedgwood's 1769 pottery factory — Konstantine · 2026-10-31
- AI memo: China to restrict exports of 23 AI technologies, plus Kimi K3.1 Pro and Genie 4 — dralandthompson · 2026-10-31
- Chinese firms offer AI 'ghost bots' that resurrect deceased loved ones from 30 seconds of footage — ArtificialOther · 2026-10-31
- SCOPE workshop proposed for ICLR 2027 asks what counts as an AI scientific discovery — mariyaivasileva · 2026-10-31