Ant open-sources LingSante-3.1-flash, a 560B healthcare model with FP8/MXFP4 weights
BerryBlush_Rogers691 · reddit · 2026-11-22
Ant Group open-sourced LingSante-3.1-flash, a 560B-parameter (A25B active) healthcare model with FP8 and MXFP4 weights on Hugging Face. Official charts claim strong medical benchmark results vs frontier models: HealthBench Professional 65.72, DiagnosisArena-MCQ 83.9, MentalHealthBench 54.64, EHR-Complex 74.40, plus general agent scores like HLE 72.22 (w/ tools) and BrowseComp 94.10, compared against DeepSeek-V4.1-Flash, GLM-5.3, Kimi-K3.1-Pro, Claude Opus 5.5, GPT-6 Astra and Gemini 4 Argon. Note several benchmarks are internal, so treat official numbers with caution.
Related event: Ant Group Open-Sources 560B Medical Model LingSante-3.1-flash(3 posts)→
More from Models
- Distillation is mostly cope: labs buy astronomical data, dev argues overhyped in local AI community — gnukeith · 2026-11-23
- Sarvam's new 105B checkpoint jumps from 42.6 to 73.9 on benchmarks, claims Flash-Lite Pareto frontier — itsOmSarraf_ · 2026-11-22
- Dev says Google models are making real progress on lossless compression as intelligence proxy — gnukeith · 2026-11-22
- Astra-6.2 'lazy' backlash blamed on user education: default manual mode vs goal mode — alexisgallagher · 2026-11-22
- Musk: Tesla has had a near pixel-perfect world model since last year — giffmana · 2026-11-22
- Gemini 4 Neon takes #1 on LiveBench 2.0 with 87.4, beating Claude and GPT by 2.5 points — demishassabis · 2026-11-22