LLMs suddenly got good at NMR since Claude Opus 4.8, says ChefNMR2 team
ZhongingAlong · x · 2026-11-03
This tweet introduces the ChefNMR2 benchmark series. The author observes that around June's Claude Opus 4.8, chatbots "very suddenly" got good at NMR — LLMs are now generally intelligent enough for scientific tasks like interpreting NMR spectra much as a chemist would.
The team therefore partnered with Edison Scientific (makers of the Kosmos AI scientist platform) to run a large-scale benchmark of frontier reasoning LLMs on NMR structure elucidation and explore integration with Kosmos.
More from Models
- Qwen 3.8 27B spontaneously uses Blender MCP to code its way through a physics problem — Balloon_Fan · 2026-11-03
- AA-PersonalityBench goes live: build your own AI personality rankings across 41 models — ArtificialAnlys · 2026-11-03
- Kimi K3.1 Pro writing style shift: fewer em dashes and asides, more bold, headings and tables — ArtificialAnlys · 2026-11-03
- AA-PersonalityBench: Kimi K3.1 Pro shows biggest personality shift vs predecessor, avg 30 percentile points — ArtificialAnlys · 2026-11-03
- I let an abliterated DeepSeek model hack me — it rooted my Samsung TV in 31 minutes — Rasmic · 2026-11-03
- antirez on why Astra can edit NN weights: RL programming training with toy NN problems — antirez · 2026-11-03