Liquid AI's LFM longevity models hit Cell cover, beating GPT-5 and Gemini-3.1-Pro with 1.2B-2.6B params
helloiamleonie · x · 2026-09-17
Liquid AI, partnering with Insilico Medicine, published a Cell cover paper on foundation models for aging biology. They released LongevityBench, a 17-task eval suite spanning clinical records, DNA methylation, transcriptomics, plasma proteomics, and genetic evidence, plus two compact fine-tuned models — LFM2-1.2B-Longevity and LFM2-2.6B-Longevity — that set SOTA on biological longevity tasks, outperforming frontier models like Gemini-3.1-Pro, GPT-5, and Claude Opus. Models and benchmark are open on Hugging Face, showing domain-tuned small models rivaling much larger general ones.
More from Models
- Teknium: Hermes will lean toward Pi-style, away from OpenClaw — Teknium · 2026-09-18
- Liquid AI's tiny longevity models top GPT-5 and Gemini in Cell-published biology benchmark — JosephJacks_ · 2026-09-18
- Investor: Anthropic's lack of image generation is a huge gap as GPT one-shots logo editing — brezina · 2026-09-18
- SPLADE's original author releases new sparse embedding model: 150M params beat models up to 1B — IgorCarron · 2026-09-18
- WisprFlow launches Canto speech model with 3.4% word error rate for real-time dictation — Scobleizer · 2026-09-18
- OpenAI reportedly close to solving Millennium Prize problem, the Hodge Conjecture — gerardsans · 2026-09-18