Anthropic IPO deck chart sparks debate: Mythos could be retrained with a fraction of the compute
Witty_County5128 · reddit · 2026-11-15
A Reddit user cites a chart from Anthropic's IPO roadshow shared by Jacob Coxon showing how quickly pre-training efficiency has improved. Coxon argues Anthropic could already retrain Mythos using far less compute than it originally took. The post asks whether this is realistic and how much further training efficiency can go, with the chart attached — pointing at the implicit compute-depreciation effect of steady pre-training gains.
Related event: Anthropic Training Efficiency Chart Sparks Debate(12 posts)→
More from Infra
- AMD's Rumored Siren Halo APU Reportedly Packs Dedicated GDDR7 On-Package Memory, Eyeing Late 2028 — Terminator857 · 2026-11-15
- GLM-5.3 on Two DGX Stations Peaks at 607.8 tok/s but 32 Concurrency Causes 453s Latency — petrusenko_max · 2026-11-15
- Pundit likens possible Huang exit to Microsoft antitrust moment as rivals circle Nvidia — TiernanRayTech · 2026-11-15
- Analyst warns of hyperscaler monopoly: NVIDIA loses pricing power, startup funding dries up — BenBajarin · 2026-11-15
- Anthropic CEO Dario Amodei explains why the company buys compute conservatively — _AustinCalvert_ · 2026-11-15
- HF Researcher: Big AI Labs Turn on Jensen Because He Fights for Open-Source — mervenoyann · 2026-11-15