Ortet Benchmarks 34 Open-Weight Health Models on 50 Tasks Across 15 Benchmarks
kchonyc · x · 2026-11-23
Health AI startup Ortet evaluated 34 recent open-weight models on 50 tasks drawn from 15 public health benchmarks, self-hosted on its Orion cluster of NVIDIA Blackwell GPUs, alongside 8 closed frontier models (GPT-6 and Claude 5.5 families) released September 2026. Notably, 25 of the 34 open models (75%) shipped within the prior six months. The team argues full-stack transparency and controllability matter most in privacy-sensitive sectors like healthcare: self-hosting lets them verify model versions, configure inference, and reproduce results. Ortet is building a health model spanning biological, clinical, and operational dimensions, and will follow up with an agentic task eval run on Orion.
More from Companies & People
- Musk touts Tesla superintelligence, chip design & Terafab as shared world model powers Autopilot and Optimus — elonmusk · 2026-11-23
- Allie Miller shares the highest-leverage AI and non-AI spending on her small team — alliekmiller · 2026-11-23
- Sequoia's Doug Leone: 30% of VCs on boards are incompetent, 40% are no-ops — aldeamartinez · 2026-11-23
- Replit ships Windows-hardened security model for desktop AI, wins Christie's — amasad · 2026-11-23
- Anthropic still leads everyone in pretraining, especially on data, researcher argues — Guodzh · 2026-11-23
- Insider claim: OpenAI never believed much in pretraining and big models through 2024-2025 — Guodzh · 2026-11-23