Smol/Gemma/Llama YOLOv11

YOLO11 is the state-of-the-art computer vision framework by Ultralytics, now frequently paired with lightweight LLMs like SmolLM, Gemma, and Llama 3.2 for advanced edge-based multimodal reasoning.

YOLO11 represents the latest leap in real-time object detection, delivering 22% fewer parameters than YOLOv8m while achieving higher mAP on the COCO dataset. Released in September 2024, it supports a full suite of vision tasks—detection, segmentation, pose estimation, and OBB—optimized for deployment on NVIDIA Jetson and edge devices. Modern AI pipelines increasingly integrate YOLO11's high-speed spatial awareness with the reasoning capabilities of small language models (SLMs) like Meta's Llama 3.2 1B or Google's Gemma 2 2B. This synergy allows developers to build systems that not only identify objects with millisecond latency but also provide natural language context and decision-making logic directly on-device.

https://docs.ultralytics.com/models/yolo11/