Ring-lite is a lightweight inference model based on the MoE architecture, developed by AntTech. The model is based on Ling-lite-1.5 and utilizes the proprietary C3PO reinforcement learning training method, achieving high performance on multiple inference benchmarks...
PosterCraft is a unified framework developed by the Hong Kong University of Science and Technology (Guangzhou) and Meituan, among other institutions, for generating high-quality, aesthetically pleasing posters. The framework abandons modular design processes and fixed predefined layouts, supporting free exploration and coherence of models...
Confucius3-Math is a mathematical model within the "Confucius Says 3" series of large-scale open-source models from NetEase Youdao. It is the first open-source reasoning model in China specifically focused on mathematics education. With 14 billion parameters, it is optimized for K-12 mathematics education scenarios and can be used in...
Hunyuan-GameCraft is a high dynamic range interactive game video generation framework jointly developed by Tencent's Hunyuan team and Huazhong University of Science and Technology. It achieves fine-grained... by unifying keyboard and mouse input into a shared camera representation space.
ImmerseGen is an innovative 3D world generation framework jointly developed by ByteDance's PICO team and Zhejiang University. Based on user-input text prompts and agent-guided asset design and arrangement, the framework generates worlds with alpha textures...
Gemini Robotics On-Device is Google DeepMind's first visual-language-motion (VLA) model that can run locally on a robot. The model possesses powerful offline operational capabilities, able to follow natural language commands to complete...
OmniAvatar is an audio-driven full-body video generation model jointly developed by Zhejiang University and Alibaba Group. Based on the input audio and text prompts, the model generates natural and realistic full-body animated videos, with character movements perfectly synchronized with the audio...
AnimaTensor is a two-dimensional image generation model jointly developed by the CagliostroLab team and TensorArt. Based on the innovative V-Prediction technology, it optimizes noise scheduling and sampling strategies by predicting the "speed" of the image generation process...
DAMO GRAPE is the world's first AI model for identifying early-stage gastric cancer based on plain CT scans, jointly launched by Zhejiang Cancer Hospital and Alibaba DAMO Academy. DAMO GRAPE breaks through the limitations of traditional imaging, using deep learning to analyze non-contrast CT scans...
Seed 1.6 is a general-purpose model series launched by ByteDance's Seed team. It integrates multimodal capabilities and supports deep inference with 256K long contexts. Seed 1.6 utilizes the sparse MoE exploration results of Seed 1.5, and undergoes pure text pre-training, multimodal hybrid...
Gemini CLI is an open-source endpoint AI programming tool from Google, using the Gemini 2.5 Pro model for free. Gemini CLI offers powerful AI capabilities, such as code understanding, file operations, command execution, and dynamic troubleshooting, assisting developers...
AlphaGenome is a new AI model from Google's DeepMind that provides a deeper understanding of the genome. The model can receive DNA sequence inputs of up to one million base pairs, predict thousands of molecular properties characterizing their regulatory activities, and assess...
Claudia is a desktop application and toolkit designed for Claude Code, enhancing the efficiency and security of AI-assisted development through an intuitive graphical interface. It supports project and session management, allowing users to easily browse and restore past sessions...
DRA-Ctrl (Dimension-Reduction Attack) is an innovative cross-modal image editing framework developed by Zhejiang University in collaboration with Ant Group and other institutions. The framework leverages the visual, temporal, spatial, and causal dimensions of video generation models...
Kwai Keye-VL is a multimodal large language model independently developed by Kuaishou. It's based on the Qwen3-8B language model and integrates a SigLIP-initialized visual encoder, supporting dynamic resolution input. The model can deeply fuse and process text, images, etc.
Twocast is an AI-powered podcast generator that supports generating podcast content in the form of two-person conversations. The generator supports multiple languages and voices, and can quickly generate 3-5 minute podcasts based on topics, links, documents, and other methods.
VLN-R1 is a new embodied intelligence framework jointly developed by the University of Hong Kong and the Shanghai Artificial Intelligence Laboratory. Based on a large-scale visual language model (LVLM), it directly converts first-person video streams into continuous navigation actions. The framework is based on Habit...
Nanonets-OCR-s (Nanonets OCR Small) is an image-to-Markdown OCR model from Nanonets that supports converting document content within images into structured Markdown format. The model can extract text...
Hunyuan-A13B is Tencent's latest open-source large language model, based on a hybrid expert (MoE) architecture, with a total of 80 billion parameters and 13 billion activation parameters. It features a lightweight design and efficient inference capabilities, requiring only one mid-to-low-end GPU...
Kling-Foley is a multimodal video audio generation model developed by Kling AI. The model takes video and text prompts as input and generates high-quality stereo audio that is semantically relevant to the video content and time-synchronized, covering sound effects...
4D-LRM (Large Space-Time Reconstruction Model) is a new 4D reconstruction model jointly developed by researchers from Adobe Research, the University of Michigan, and other institutions. The model can reconstruct data based on sparse input views and arbitrary...
FilMaster is an AI filmmaking system jointly developed by the University of Hong Kong, Kuaishou Technology, Microsoft Research, and Tsinghua University. It combines filmmaking principles with generative artificial intelligence to achieve fully automated production from script to final film...
MultiAgentPPT is a multi-agent presentation generation system based on the A2A (Ask-to-Answer), MCP (Multi-agent Control Protocol), and ADK (Agent Development Kit) architecture. MultiA...
Qwen VLo is a multimodal unified understanding and generative model developed by the Tongyi Qianwen team. It represents a comprehensive upgrade from the large multimodal model, enabling it to "see" the world and perform high-quality re-creation based on that understanding, achieving a transformation from perception to...