SenseNova-Vision: A 7B Open Model Unifying Segmentation, Depth, and 3D Reconstruction
SandyL925 · reddit · 2026-08-13
SenseNova has open-sourced the 7B vision model SenseNova-Vision (Apache 2.0). Using a Mixture-of-Transformers (MoT) architecture, it treats nearly all computer vision tasks as a single generation problem.
- Core Capabilities: Without task-specific prediction heads, it handles object detection, keypoints, OCR, various segmentations (binary, instance, semantic), and depth/surface normal estimation using a single set of weights.
- Key Highlight: Supports multi-view 3D reconstruction and camera pose estimation. Typically requiring specialized tools like COLMAP, this model accomplishes it with a single prompt.
- Training & Deployment: Trained on 50M instruction-response pairs. A web demo and Hugging Face weights are available. However, hardware requirements are steep: the full web demo needs at least a single 80GB GPU, and benchmarking requires 8x80GB GPUs.
More from Models
- Devs say LLM coding mistakes are a different class: overcomplication, inferred limits, worse at higher reasoning — Mu5_ · 2026-10-16
- ChatGPT UI screenshot reveals Wallet coming soon plus Email, Calendar, Health and Finances integrations — jxnlco · 2026-10-16
- Why the next OpenAI and Anthropic models may still disappoint: no native audio or video input — imjustnewatai · 2026-10-16
- Thousands Hit by ChatGPT Standard Voice Mode Regression: Wrong Voice, Interruptions — shaunralston · 2026-10-16
- Gemini Argon 4 Quota Exhausted, Reset Countdown Shows Nearly 7 Days — ssh4net · 2026-10-16
- AINews: Meta ships 816B Muse Spark Open; Goodfire argues neural nets are already clean — Latent Space · 2026-10-16