Poland's 9B-parameter Thalamus Reasoning tops LongevityBench, beating 29 models
Dr_Singularity · x · 2026-11-23
Warsaw-based Ingenix.ai reports its Thalamus Reasoning model ranked #1 of 29 on LongevityBench, an aging-research benchmark: first on 11 of 14 tasks, mean rank 1.42 (next best 4.42; best frontier model Claude Opus 4.6 at 9.85), with 3.36-year methylation age error.
The 9B-parameter model skips biology text and reads biological data directly: 110,000 methylation arrays, 160 Olink proteomics studies, plus GTEx, TCGA and DepMap. The company calls it a Biological Reasoning Engine using Modality Fusion, targets clinical trial efficacy/safety decisions, and is open-sourcing a researcher version.
More from Models
- 55-Page Riemann Hypothesis Proof Guide, De-Formalized from Lean with Gemini and GPT — chaumian · 2026-11-23
- Sarvam open-sources its latest 105B model; founder says scaling agentic RL 'is not for the faint hearted' — itsOmSarraf_ · 2026-11-23
- Gemini 4 Neon xhigh costs 8.6x more but adds just 1-4 points on most benchmarks — jeremyphoward · 2026-11-23
- Dev: Gemini 4 Neon rewrote my training kernels and supervised runs; rumors point to bigger models — xennygrimmato_ · 2026-11-23
- Ant Group open-sources LingSante-3.1-air-VL, a 124B MoE medical multimodal model — jacek2023 · 2026-11-23
- Inkling 2 Mini: a 26B local model that runs on laptops, and someone rebuilt Apple Notes around it — simonguozirui · 2026-11-23