LlamaV-o1 is a novel multimodal visual reasoning model proposed by institutions such as the Mohammed bin Zayed University for Artificial Intelligence in the UAE, enhancing the stepwise visual reasoning capabilities of large language models. It introduces the VRC-Benc benchmark for visual reasoning chains...
Kokoro-TTS is a lightweight text-to-speech (TTS) model developed by hexgrad, featuring 82 million parameters. It's based on a hybrid architecture of StyleTTS 2 and ISTFTNet, employing a pure decoder design and avoiding the use of a diffusion model...
KHOJ is an open-source, personalized AI assistant that helps users integrate and retrieve knowledge. It connects to users' online and local documents, such as PDFs, Markdown, plain text, GitHub and Notion files, and uses semantic search to quickly...
Luma Ray2 is the latest video generation model from Luma AI. Trained on Luma's new multimodal architecture, Ray2 exhibits advanced capabilities, boasting 10 times the computational power of Ray1. It can generate fast and coherent videos...
RAIN (Real-time Animation Of Infinite Video Stream) is an innovative real-time animation solution that enables real-time animation of infinite video streams using consumer-grade hardware, such as a single RTX 4090 GPU. Core...
Pipecat is an open-source Python framework focused on building speech and multimodal dialogue agents. Based on built-in speech recognition, text-to-speech (TTS), and dialogue processing capabilities, it simplifies the complex coordination, network transmission, and audio processing of AI services...
RealtimeSTT is an open-source real-time speech-to-text library designed for low-latency applications. It features powerful speech activity detection capabilities, automatically recognizing the start and end of speech, and using WebRTCVAD and SiloVAD for accurate detection...
Step R-mini (full name Step Reasoner mini) is a reasoning model launched by Step Star. It is the first reasoning model in the Step series of models, excelling in proactive planning, trial and error, and reflection, based on slow thinking and repeated verification...
GLM-Realtime is a brand-new end-to-end multimodal model launched by Zhipu, featuring low-latency video understanding and voice interaction capabilities. It also incorporates a singing function, allowing the large model to showcase its singing abilities during conversations. The model supports up to 2 minutes of...
CogView-3-Flash is the first free AI image generation model launched by Zhipu. It can generate high-aesthetic-score images based on text descriptions, supports multiple resolutions, and meets the needs of professional fields. The model boasts creative diversity and is based on...
MangaNinja is a reference-based line art coloring method that offers precise matching and fine-grained control. Through an innovative patch rearrangement module and point-driven control scheme, it improves coloring accuracy and image quality. It can handle...
AutoMouser is a Chrome extension that intelligently tracks user interactions and automatically generates Selenium test code based on OpenAI's GPT model. It records user browser actions such as clicks, drags, and hovers...
CogVideoX-Flash is the first free AI video generation model launched by Zhipu. Inheriting the self-developed end-to-end video understanding model of CogVideoX, it possesses powerful video generation capabilities. CogVideoX-Flash supports text-based video, and more...
Mini-InternVL is a "mini" version of the Shusheng·Wanxiang large model. It is a lightweight, multimodal, large-scale language model series jointly launched by the Shanghai AI Lab, Tsinghua University, Nanjing University, and other institutions. It includes three parameter versions: 1B, 2B, and 4B...
Weebo is a real-time voice chatbot powered by Whisper Small, Llama 3.2, and Kokoro-82M technologies. It uses speech recognition and generation technology to engage in natural and fluent conversations with users, providing real-time voice interaction...
ParGo is an innovative multimodal large language model connector proposed by ByteDance in collaboration with Sun Yat-sen University. It improves the alignment of visual and linguistic modalities in multimodal large language models (MLLMs). This is achieved by combining local tokens and global tokens...
MatterGen is an innovative generative model from Microsoft, specifically designed for inorganic materials. Based on a unique diffusion process, it progressively refines atomic types, coordinates, and periodic lattices to generate stable and diverse inorganic materials spanning the periodic table...
XMusic is an AI-powered universal music framework independently developed by Tencent Multimedia Lab. Users only need to upload any content such as videos, images, text, tags, or humming, and XMusic can generate high-quality music with controllable emotion, style, and rhythm...
Seaweed APT is an adversarial post-training model launched by ByteDance, capable of one-stop generation of images and videos. Seaweed APT is based on a pre-trained diffusion model, directly applying it to real-world data...
OmAgent is a multimodal language proxy framework jointly open-sourced by Om AI and the Binjiang Research Institute of Zhejiang University. It simplifies the development of intelligent agents on device-side. OmAgent supports reusable proxy components, helping developers build complex multimodal...
The T2A-01-HD is a new voice model launched by Conch AI's overseas version. It supports voice cloning, accurately cloning a voice with just 10 seconds of audio while preserving its emotional nuances. The model features an intelligent emotion system that can capture subtle emotional differences in speech...
Uni-AdaFocus is a general-purpose, high-efficiency video understanding framework developed by a research team from the Department of Automation at Tsinghua University. The framework uses an adaptive focusing mechanism to dynamically adjust the allocation of computing resources, enabling efficient processing of video content.
AnyStory is an innovative text-to-image generation framework developed by Alibaba's Tongyi Lab, enabling high-fidelity personalized image generation for single and multiple subjects. It models subject personalization using an "encoding-routing" method. ...
SHMT (Self-supervised Hierarchical Makeup Transfer) is an advanced self-supervised makeup transfer technology jointly developed by Alibaba DAMO Academy and Wuhan University of Technology, among other institutions. The technology utilizes a latent diffusion model to achieve...