dots.llm1 is a medium-sized Mixture of Experts (MoE) text model open-sourced by Xiaohongshu's hi lab, with 142 billion parameters and 14 billion activation parameters. The model was tested on 11.2TB of high-quality token data...
MonkeyOCR is a document parsing model jointly developed by Huazhong University of Science and Technology and Kingsoft Office. The model supports the efficient conversion of unstructured document content into structured information. Based on accurate layout analysis, ...
Playmate is a face animation generation framework developed by the Guangzhou Quwan Technology team. Based on a 3D implicit space guided diffusion model, the framework uses a two-stage training process to precisely control facial expressions and head postures based on audio and commands, generating...
Time-R1 is a language model based on 3B parameters developed by a research team at the University of Illinois at Urbana-Champaign. Through a unique three-stage reinforcement learning training method, it has achieved a significant breakthrough in time reasoning ability.
MiniMax-Remover is a novel video target removal method that addresses common issues in existing technologies, such as illusory objects, visual artifacts, and slow inference speed. It employs a two-stage approach: the first stage is based on a simplified DiT framework...
MAS-Zero is a multi-agent system (MAS) design framework from Salesforce that automatically designs and optimizes MAS without human supervision. The framework is based on meta-level design, dynamically adjusting each problem instance during inference...
MultiTalk is a novel audio-driven multi-person dialogue video generation framework jointly developed by Sun Yat-sen University Shenzhen Campus, Meituan, and Hong Kong University of Science and Technology. The framework generates videos containing human dialogue based on multi-channel audio input, reference images, and text prompts...
SmolVLA is a lightweight, open-source vision-language-action (VLA) model from Hugging Face, designed for cost-effective robotics. It boasts 450 million parameters, is compact, and can run on a CPU, requiring only a single consumer-grade GPU...
ContentV is an open-source image-based video model framework with 8 billion parameters developed by ByteDance. It replaces the 2D-VAE of Stable Diffusion 3.5 Large with a 3D-VAE and introduces 3D positional encoding, enabling image models to quickly acquire video generation capabilities.
PartCrafter is an advanced 3D generative model capable of generating multiple semantically well-defined and geometrically diverse 3D meshes from a single RGB image. It combines latent space representations of each 3D part and uses a hierarchical attention mechanism to...
MaskSearch is a novel general-purpose pre-training framework launched by Alibaba Tongyi Labs to enhance the agent search capabilities of Large Language Models (LLMs). Through the Retrieval Augmented Mask Prediction (RAMP) task, it allows the model to perform search on input text...
InftyThink is an innovative paradigm for large-scale model inference, overcoming the limitations of traditional models in long-duration inference tasks. It breaks down complex inference processes into multiple short segments through segmented iteration, generating an intermediate summary after each segment...
o3-pro is a powerful inference model from OpenAI. As an upgraded version of o3, it excels in handling complex problems and providing accurate answers, especially in scenarios such as scientific research, programming, education, and writing. o3...
Magistral is an inference model developed by Mistral AI, focusing on transparent, multilingual, and domain-specific reasoning capabilities. The model includes Magistral Small (open-source version) and Magistral Medium (enterprise version)...
Vui is a lightweight voice dialogue model open-sourced by the Fluxions-AI team, based on the LLaMA architecture. The model has undergone 40,000 hours of dialogue training and can simulate interjections, laughter, and pauses in real conversations, providing an immersive interactive experience...
Krea 1 is an AI image generation model from Krea AI that addresses the "AI aesthetics" problem in traditional AI image generation. The model can generate highly realistic, textured images, supports multiple artistic styles, and provides style references...
Seedance 1.0 is a basic video generation model launched by ByteDance's Seed team. The model supports text and image input, can generate 1080p high-quality videos with seamless multi-camera switching, possesses native multi-camera storytelling capabilities, and can...
Doubao-Seed-1.6 is a multimodal deep thinking model launched by ByteDance. The model supports three thinking modes: auto, thinking, and non-thinking, and can process text, images, videos, and more...
Reor is an open-source, localized AI-powered personal knowledge management application. It supports automatic linking of related notes, semantic search, and Q&A functionality. Users edit notes using a Markdown editor similar to Obsidian. Reor is based on Ol...
V-JEPA 2 is a world-class model launched by Meta AI, enabling the understanding, prediction, and planning of the physical world based on video data. V-JEPA 2 utilizes the Joint Embedded Prediction Architecture (JEPA) with 1.2 billion parameters, based on self-supervised learning...
HistAgent is an AI assistant system jointly developed by the Princeton University AI Lab and the Department of History at Fudan University, designed specifically for historical research. It addresses multimodal information processing, cross-linguistic analysis, and complex reasoning in historical research...
SeedVR2 is a new single-step video restoration (VR) model launched by ByteDance, based on a diffusion model and Adversarial Post-Training (APT) techniques. The model is based on...
Next-Frame Diffusion (NFD) is an autoregressive video generation model jointly developed by Peking University and Microsoft Research. It combines the high-fidelity generation capabilities of diffusion models with the causality and controllability of autoregressive models. Next-Frame...
LLIA (Low-Latency Interactive Avatars) is a real-time audio-driven portrait video generation framework developed by Meituan, based on a diffusion model. The framework generates virtual avatars based on audio input and supports low latency...