BLIP3-o is an innovative multimodal model developed by Salesforce Research and other institutions, combining the inference and instruction-following capabilities of autoregressive models with the powerful generative capabilities of diffusion models. The model is based on CLIP images with rich diffusion semantics...
ShotAdapter is a framework developed by Adobe in collaboration with UIUC for text-to-multi-camera video generation. Based on a fine-tuned pre-trained text-to-video model, it introduces transition markers and local attention masking strategies to achieve multi-camera video...
Index-AniSora is an advanced anime video generation model open-sourced by Bilibili. The model supports one-click generation of videos in various anime styles, covering anime clips, domestically produced animation, VTuber content, and more. The model contains over 10 million...
LightLab is an image light source control model based on a diffusion model, developed by Google and other institutions. It enables fine-grained parameterized control of light sources within a single image. Users can adjust the intensity and color of the light source, insert virtual light sources, ...
WorldPM (World Preference Modeling) is a series of preference modeling models developed by Alibaba Group's Qwen team and Fudan University. It reveals the scalability of preference models through large-scale training. The model is based on 15 million...
SketchVideo is a sketch-based video generation and editing framework developed by the University of Chinese Academy of Sciences, the Hong Kong University of Science and Technology, and the Kuaishou Keling team. It allows users to draw sketches on keyframes and combine them with text prompts to achieve spatial layout of the video...
Custom-SVG is a two-stage SVG customization framework developed by Adobe and City University of Hong Kong. It generates high-quality SVG graphics with customized styles based on text prompts. The framework introduces a two-stage generation workflow...
YuLan-OneSim is a new type of social simulator developed by the RUC-GSAI team at Renmin University of China's Gaoling AI Institute. Based on Large Language Model (LLM) agents, it simulates human social behavior without requiring programming to build the model...
MathModelAgent is an agent specifically designed for mathematical modeling, capable of automating the entire process from problem analysis to paper generation. Based on multi-agent collaboration, it automatically builds mathematical models, writes code, verifies results, and generates formatted documents...
Steamer-I2V is an image-to-video generation model developed by Baidu's Steamer team. It demonstrates exceptional visual generation capabilities by transforming static images into dynamic videos. The model has achieved high scores on the internationally recognized VBench video generation benchmark...
ScaleMCP is a tool selection methodology developed by PwC that dynamically equips Large Language Model (LLM) agents with Model Context Protocol (MCP) tools. It is based on an automatically synchronized tool storage system and MCP server...
WeClone is a one-stop solution for creating digital avatars from chat logs. Based on chat log-based fine-tuning of Large Language Models (LLM), it gives the model a specific style and binds it to chatbots such as WeChat, QQ, and Telegram...
Agent Squad is a lightweight, open-source multi-agent framework used to coordinate multiple AI agents to handle complex dialogues. Agent Squad supports both Python and TypeScript, and dynamically routes data based on user input...
Magentic-UI is an open-source research prototype from Microsoft used to explore human-participatory AI agent systems. Magentic-UI is a human-centered AI agent capable of collaborating with users to complete complex web tasks, such as browsing...
EvoAgentX is an open-source AI agent self-evolution framework that uses evolutionary algorithms to automate the generation and optimization of multi-agent systems. It can automatically generate workflows based on target descriptions and improve performance through iterative optimization.
NLWeb is an open-source project launched by Microsoft that simplifies the creation of natural language interfaces for websites, enabling any website to become an AI-driven application. NLWeb uses semi-structured data such as Schema.org and RSS, combined with LLM...
Doubao Voice Podcast Model is a voice podcast model launched by Volcano Engine, a subsidiary of ByteDance. Based on a streaming model, it can convert text into a two-person conversational podcast in seconds, featuring low cost, high efficiency, and strong interactivity. ...
Veo 3 is a next-generation video generation model released at Google I/O. Veo 3 is Google's first model capable of generating background sound effects for videos. It can synthesize images and add corresponding sound effects to scenes such as birdsong and street traffic.
II-Agent is an open-source agent framework that simplifies and enhances workflows across multiple domains by interacting with large language models (LLMs). It possesses several core functionalities, including research and fact-checking, content generation, data analysis, etc.
News Agents is a terminal-based news aggregation and summarization system. It uses Amazon Q CLI as the agent framework, Model Context Protocol (MCP) to parse RSS news sources, and leverages tmux to achieve terminal segmentation and multi-tasking...
Imagen 4 is Google's latest AI model for image generation. It supports image generation up to 2K resolution, delivering lifelike details and clearly rendering complex fabric textures, water droplet refraction, and animal fur textures. Regarding text rendering...
Pixel3DMM is a single-image 3D face reconstruction framework jointly developed by the Technical University of Munich, University College London, and Synthesia. Based on the DINO model, the framework introduces a specialized prediction head to accurately reconstruct faces from a single RGB image...
AutoBE is an AI-driven backend server code generation tool that automatically generates high-quality backend code based on user-described requirements. Built on a technology stack including TypeScript, NestJS, Prisma, and Postgres, it emphasizes...
Gemma 3n is an edge-based multimodal AI model introduced at Google I/O. Based on the Gemini Nano architecture, it employs a layer-by-layer embedding technique to compress memory usage to the level of a 2-4B parameter model. The number of model parameters is...