Anthropic internal models may beat human record on nanogpt, researcher says
eliebakouch · x · 2026-11-15
Anthropic-affiliated researcher eliebakouch says his nanogpt experiment results depend heavily on environment definition: most gains come from hyperparameter tuning that actually incentivizes the model in the right direction. He notes newer models like Opus 5.5 already beat the human record in his setup, with a blog post planned.
He guesses current internal Anthropic models (or non-nerfed versions) would do even better, but doubts labs will release proper benchmarks for this soon.
More from Models
- Gemini math lead reveals inside story: evals, post-training recipes, scaled runs — thtrieu_ · 2026-11-15
- Blogger saves $53 in 43 hours after switching from Claude Code to Gemini 4 Carbon — mazzaTalk · 2026-11-15
- SatBench: 63% of lab-cited AI benchmarks already saturated, and solving is speeding up — emollick · 2026-11-15
- LFM2.5-24B-A2B praised for local agent use: 37 tok/s on an AMD 780M iGPU — jeremyckahn · 2026-11-15
- Will Depue's Blackwell 'forensic evidence' claim on DeepSeek hits snag: V4.1-Flash not cracked — teortaxesTex · 2026-11-15
- nanoGPT Speedrun Gains Mostly From Environment Design; Newer Models Beat Human Record — eliebakouch · 2026-11-15