inclusionAI open-sources Ling-3.0-flash-Sante, a 124B-parameter healthcare model activating just 5.1B per token
jacek2023 · reddit · 2026-10-22
inclusionAI has released Ling-3.0-flash-Sante, a healthcare-specialized open-source model built on Ling-3.0-flash, available on Hugging Face and ModelScope.
- Scale: 124B total parameters with only 5.1B active per token, using a hybrid-linear MoE architecture combining Kimi Delta Attention, gated MLA, and sparse MoE layers
- Context: trained to 256K tokens for long-document analysis and evidence synthesis
- Positioning: built for medical reasoning, healthcare research, deep research, tool-assisted workflows, and multi-agent collaboration, with thinking mode on by default
- Benchmarks: strong results on professional healthcare tasks, medical knowledge, and healthcare safety benchmarks
The team stresses it is a research model, not a medical device, and high-stakes use requires qualified human review.
Related event: Ant Group Open-Sources Medical MoE Model Ling-3.0-flash-Sante(5 posts)→
More from Models
- Microsoft's step-by-step diagnostic AI lands Science cover, code released — erichorvitz · 2026-10-23
- Sophont's Holon-S1 predicts genes and biomarkers from routine pathology slides — iScienceLuvr · 2026-10-23
- Doc parsing lite mode scores 75.1 vs Azure Read's 42.3 on OmniParseBench, dominates non-Latin scripts — VikParuchuri · 2026-10-23
- Datalab open-sources Indra OCR: a 1M-parameter model that beats PaddleOCR on CPU — VikParuchuri · 2026-10-23
- The inside story behind OpenAI's million-dollar math proof — scientificamerican · 2026-10-23
- Anthropic's anti-distillation defense cuts attacks 90-99%, but expert data is still for sale — johnkorrigan · 2026-10-23