Hardware
Introducing NV-Reason-CT Open 3D CT VLM for Radiologist Chain-of-Thought Reasoning
NV‑Reason‑CT is a purpose‑built vision‑language model for volumetric CT that uses a full 3‑D Vision Transformer encoder coupled with a Qwen 3.5‑4B language model to generate radiologist‑style chain‑of
NV‑Reason‑CT is a purpose‑built vision‑language model for volumetric CT that uses a full 3‑D Vision Transformer encoder coupled with a Qwen 3.5‑4B language model to generate radiologist‑style chain‑of‑thought reasoning over anatomical regions. It achieves state‑of‑the‑art results on the CT‑RATE benchmark, attaining a Macro‑F1 of 0.614 and Macro‑AUROC of 0.871, surpassing existing 3‑D contrastive and fused 2‑D/3‑D baselines. NIH radiologists have confirmed the clinical plausibility and auditability of its structured reports, and the model is released under an open research foundation that supports post‑training for specialized CT applications within NVIDIA’s Medical AI ecosystem.
Related
- Generate Trajectories, Reasoning Traces, and Auto-Labels with NVIDIA Alpamayo 2 Super
- Run Step 3.7 Flash on NVIDIA GPUs with Enterprise-Ready Multimodal AI
- AI Model Co-Design: Hardware-Friendly LLM Design
Source: NVIDIA Developer | 2026-09-23