Insilico Medicine releases open Longevity AI toolkit and benchmark in Cell study
Insilico Medicine announced that a study published in Cell introduces LongevityBench, an open benchmark covering 17 tasks across clinical data, genetics, epigenetics, transcriptomics and proteomics. The paper evaluated 18 frontier AI systems from OpenAI, Google, Anthropic, xAI, DeepSeek and Moonshot AI, finding Gemini 3.1 Pro (score 8.2) and Claude Opus‑4.6 (9.2) as the top large‑scale models. To address the gap between frontier resources and practical research, Insilico fine‑tuned a family of five compact Longevity‑LLMs ranging from 0.6 B to 9 B parameters using its MMAI Gym and architectures from Liquid AI (LFM2) and Alibaba (Qwen3/3.5). The best compact model, L‑Qwen3.5‑9B, achieved an…
Key points
- Insilico released LongevityBench benchmark (17 tasks across five biodata domains) and five compact Longevity‑LLMs (0.6‑9 B parameters).
- Compact model L‑Qwen3.5‑9B achieved a 4.4 aggregate rank, outperforming frontier models on age‑prediction tasks.
- Open‑source Longevity Claw agent identified 328 candidate aging‑intervention genes, with up to 5.6‑fold enrichment and validation of KDM1A.
Sources
- Insilico Medicine Releases Open Longevity AI Toolkit in Cell Study Unite.AI
- How fast are you ageing? Ask AI Nature Machine Learning
This story is more than 2 months old and kept in the archive in short form. The summary and key points were generated by Digest AI's editorial model from the linked sources, which are the record. For today's news, see the briefing or the Research section.