New scaling curve has half the slope: 10,000x compute for 20%-to-80%, but bigger generational jumps
tobyordoxford · x · 2026-09-14
tobyordoxford says he'll write more on a recent inference-scaling graph, highlighting two features: 1) the slope is half that of o1/o3 on AIME — needing 10,000x compute to go from 20% to 80%; 2) yet the jump between model generations is much better than in the o1→o3→GPT-5 era. Inference scaling's marginal returns worsen while per-generation baseline gains grow.
Related event: New Model's Scaling Curve Halves in Slope but Shows Bigger Jumps(2 posts)→
More from Models
- David Bellamy clarifies his experiment used K2 Horizon, an open-weights 375B LLM — JeremyNguyenPhD · 2026-09-14
- Swift-Qwen3.8-27b, a token-efficient reasoning Qwen finetune, trends on Hugging Face — ukisai · 2026-09-14
- GPT-6 Astra hands-on: composes first, orchestrates later, and reportedly outshines Fable and Sol — paw_lean · 2026-09-14
- Researcher posts proof he both synthesized viruses and trained a 375B open-weight LLM — ethanCaballero · 2026-09-14
- Toby Ord: 10x more RLVR compute cuts tokens-to-target ~3x; gains may be math-specific — tobyordoxford · 2026-09-14
- Fudan NLP paper explains why max reasoning settings can backfire on SWE benchmarks — karminski3 · 2026-09-14