Fengyu is the world's first full-chain space weather model, jointly launched by the China Meteorological Administration, Nanchang University, and Huawei. Based on an innovative chain-based training structure, the model combines satellite observation data with numerical model data to achieve solar wind...
RepText is a multilingual visual text rendering framework developed by Shakker Labs and Liblib AI. It achieves high-quality text rendering by copying glyphs rather than understanding the text content. The framework is based on pre-trained monolingual text-to-image...
Xiaomi MiMo is Xiaomi's first open-source reasoning model, designed to improve performance on complex reasoning tasks. The model leverages combined pre-training and post-training, mining a large amount of rich reasoning corpus and employing innovative...
F-Lite is a 10-parameter text-based image model open-sourced by the Freepik team in collaboration with FAL. It is trained on Freepik's proprietary 80M dataset and supports commercial use. F-Lite uses T5-XXL as the text encoder, based on extracting the 17th...
ACI.dev is an open-source infrastructure platform designed specifically for AI agents, providing robust tool invocation support. It integrates over 600 pre-built tools, covering common applications such as Gmail, Slack, and Notion, and supports...
URM (Universal Recommendation Model) is a global knowledge model launched by Alibaba's Alimama platform. Based on knowledge injection and information alignment, it combines the general knowledge of LLM with the professional knowledge of the e-commerce field to solve the problems of traditional LLM...
DeepSeek-Prover-V2 is an open-source, large-scale language model focused on mathematical reasoning, developed by the DeepSeek team. It includes two versions: DeepSeek-Prover-V2-671B and DeepSeek-Prover-V2-7B,...
Rowboat is an open-source, low-code AI IDE focused on building multi-agent assistants for multi-cloud platforms (MCPs). Through a visual interface and AI-assisted development features, it helps users quickly design, configure, and test intelligent agents...
Qwen2.5-Omni-3B is a lightweight multimodal AI model launched by Alibaba's Qwen team. It's a simplified version of Qwen2.5-Omni-7B, designed specifically for consumer hardware, and supports various input methods including text, audio, images, and video...
TesserAct is an innovative 4D embodied world model that predicts the dynamic evolution of 3D scenes over time and responds to the actions of embodied agents. It learns by training on RGB-DN (RGB, depth, and normal) video data, surpassing traditional...
VoltAgent is an open-source TypeScript framework used for building and orchestrating AI agents. VoltAgent provides the infrastructure and tools for building AI applications, simplifying interaction with Large Language Models (LLMs), state management, ...
Aero-1-Audio is a lightweight audio model developed by LMMs-Lab, built on Qwen-2.5-1.5B, containing only 150 million parameters. Designed specifically for long audio processing, it can support continuous audio input up to 15 minutes, without...
DianJin-R1 is a large-scale inference enhancement model for the financial field, jointly developed by Alibaba Cloud's Tongyi Dianjin team and Soochow University. Designed specifically for financial tasks, it integrates advanced technology and comprehensive data support. The model enhances supervised inference...
X-Fusion is a multimodal fusion framework jointly proposed by UCLA, UC Wisconsin-Madison, and Adobe Research. It extends pre-trained large language models (LLMs) to multimodal tasks...
Phi-4-reasoning is a 14 billion parameter inference model from Microsoft, designed for complex inference tasks. It is trained using supervised fine-tuning (SFT) and utilizes high-quality inference generated by OpenAI's o3-mini model...
Paper2Coder, developed by the Korea Advanced Institute of Science and Technology (KAIST), is a multi-agent large language model (LLM) system that automatically converts scientific papers in the field of machine learning into codebases. Based on three phases—planning, analysis, and code generation—it...
HoloTime is a panoramic 4D scene generation framework developed by Peking University Shenzhen Graduate School and Pengcheng Laboratory. Based on a video diffusion model, it transforms a single panoramic image into a panoramic video with realistic dynamic effects, which is then further reconstructed into...
T2I-R1 is a novel text-to-image generation model jointly developed by the Chinese University of Hong Kong and the Shanghai AI Lab. Based on a two-layer reasoning mechanism, semantic-level chained reasoning (CoT) and token-level CoT, it achieves high-level image planning and low-level...
Cobra (Efficient Line Art Colorization with BRoAder References) is a comic line art coloring framework developed by Tsinghua University, the Chinese University of Hong Kong, and Tencent ARC Labs. It is designed for high-precision, high-efficiency, and...
KeySync is a high-resolution lip-sync framework developed by Imperial College London and the University of Wrocław, supporting the alignment of input audio with lip movements in video. KeySync is based on a two-stage framework, first generating key...
Omni Reference is a versatile reference feature introduced by Midjourney, allowing users to embed specific people, objects, or scenes from reference images into the generated image. Omni Reference is compatible with Midjourney V7...
3DV-TON (Textured 3D-Guided Consistent Video Try-on via Diffusion Models) is a video virtual try-on based on diffusion models, jointly developed by Alibaba DAMO Academy, Lakeside Lab, and Zhejiang University...
ACE-Step is an open-source music generation model jointly launched by ACE Studio and StepFun. Based on an innovative architecture, it enables efficient, coherent, and controllable music creation. ACE-Step combines diffusion models, deep compression...
ReasonIR-8B is a model developed by Meta AI specifically designed for reasoning-intensive retrieval tasks. Trained on LLaMA3.1-8B, it employs a dual-encoder architecture, encoding queries and documents into embedding vectors separately, and then using cosine multiplication...