ColModernVBERT is a late interaction version of ModernVBERT, specifically fine - tuned for the visual document retrieval task and is the best - performing model on this task. It is a compact vision - language encoder with 250 million parameters, achieving performance close to that of models with 10 times more parameters in visual document benchmark tests, while having a considerable inference speed on the CPU.

MultimodalSafetensorsEnglish

Isaac 0.1 is the first perceptual language model launched by PerceptronAI. It has 2 billion parameters and is designed for real-world applications. This model sets a new standard in efficiency, with performance reaching or exceeding that of models more than 50 times larger. It has powerful visual understanding and spatial reasoning capabilities.

MultimodalTransformers

CHURRO is an open-weight visual language model with 3 billion parameters, specifically designed for historical document transcription. It can recognize handwritten and printed text spanning 22 centuries and 46 language clusters, including historical and extinct languages, achieving higher accuracy than large commercial models at a significantly reduced cost.

MultimodalSafetensors

FastVLM is an efficient visual encoding scheme for vision-language models. It significantly reduces the encoding time of high-resolution images through the innovative FastViTHD hybrid visual encoder, improving model performance and response speed, and performs excellently in multiple vision-language understanding benchmark tests.

MultimodalSafetensors

FastVLM is an efficient visual coding scheme designed specifically for vision-language models. Through the innovative hybrid visual encoder FastViTHD, it significantly reduces the encoding time of high-resolution images and improves the model's response speed and performance. This model performs excellently in multiple visual question-answering and document understanding benchmark tests.

MultimodalSafetensors

GLM-4.5V-AWQ-4bit is a quantized multimodal model built on Zhipu AI's next-generation flagship text foundation model. Optimized by AWQ-4bit quantization technology, it significantly reduces the computational resource requirements while maintaining excellent performance. This model achieves SOTA performance among models of the same scale in 42 public visual language benchmark tests and has powerful visual reasoning capabilities.

MultimodalTransformersMultiple Languages

DINOv3 is a series of general-purpose visual foundation models that can outperform specialized state-of-the-art techniques on a wide range of visual tasks without fine-tuning. This model can generate high-quality dense features and performs excellently in various visual tasks, significantly surpassing previous self-supervised and weakly-supervised foundation models.

Computer VisionTransformersEnglish

DINOv3 is a series of general visual foundation models developed by Meta AI. Without fine-tuning, it can outperform specialized state-of-the-art models in a wide range of visual tasks. This model uses self-supervised learning to generate high-quality dense features and performs excellently in various tasks such as image classification, segmentation, and depth estimation.

Computer VisionTransformersEnglish

DINOv3 is a versatile visual foundation model developed by Meta AI. It can outperform specialized models in a wide range of visual tasks without fine-tuning. This model can generate high-quality dense features and performs excellently in various visual tasks, significantly surpassing previous self-supervised and weakly supervised foundation models.

Computer VisionTransformersEnglish

DINOv3 is a series of general visual foundation models developed by Meta AI. It can outperform specialized advanced models in various visual tasks without fine-tuning. The model adopts the Vision Transformer architecture and is pre-trained on 1.689 billion web images. It can generate high-quality dense features and performs excellently in tasks such as image classification, segmentation, and retrieval.

Computer VisionTransformersEnglish

DINOv3 is a series of general-purpose visual foundation models that can outperform specialized state-of-the-art models on a wide range of visual tasks without fine-tuning. The model uses self-supervised learning to generate high-quality dense features and performs excellently in various visual tasks, significantly surpassing previous self-supervised and weakly supervised foundation models.

Computer VisionTransformersEnglish

DINOv3 is a series of general visual foundation models that can outperform specialized state-of-the-art techniques in a wide range of visual tasks without fine-tuning. The model generates high-quality dense features through self-supervised learning and performs excellently in various visual tasks, significantly surpassing previous self-supervised and weakly supervised foundation models.

Computer VisionTransformersEnglish

This project focuses on visual localization and reasoning with traceable evidence enhancement. It conducts research using specific base models and datasets, which has important academic and application value.

MultimodalTransformers

Llama3-MS-CLIP is the first visual language model in the CLIP family that can understand multispectral images. It is trained on one million image-text pairs from the SSL4EO-S12-v1.1 dataset and the generated descriptions, and outperforms other RGB-based models in most benchmark tests.

A specialized license plate detection model fine-tuned from Gemma 3 4B, demonstrating visual-language models can achieve localization without object detection pre-training

MultimodalTransformers

VisualPRM-8B-v1.1 is an advanced multimodal process reward model with 8 billion parameters, which enhances the reasoning ability of large multimodal language models through the Best-of-N evaluation strategy.

MultimodalTransformersmultilingual

LSNet is a family of lightweight vision models inspired by the dynamic multi-scale capabilities of the human visual system, achieving a balance between performance and efficiency across various vision tasks.

Computer VisionTimm

ZenCtrl is an intelligent agent that automates personalized visual content creation workflows, capable of performing various design tasks and training custom models.

A multilingual visual retriever combining Qwen2 vision model with stella_en_1.5B_v5, ranked first among models with ≤2B parameters in Vidore benchmark

MultimodalSafetensorsMultiple Languages

DeepSeer is a vision-language model developed based on the DeepSeek-R1 model, supporting chain-of-thought reasoning and trained through dialogue templates for visual models.

MultimodalTransformers