Higgs Audio V2 is an open-source speech model developed by Li Mu and his team at Boson AI. Trained on over 10 million hours of audio data, it features multilingual dialogue generation, automatic prosody adjustment, voice cloning, and singing capabilities...
Qwen3-235B-A22B-Thinking-2507 is Alibaba's most powerful open-source inference model globally. Based on a sparse hybrid expert (MoE) architecture with 235 billion parameters, it activates 22 billion parameters per activation and has 94 Transformer layers...
Coze Studio is an open-source AI agent development platform from ByteDance, helping developers quickly build, deploy, and manage AI agents. It provides a one-stop development environment, supporting Prompt, RAG, Plugins, W...
Coze Loop is an open-source management platform developed and operated by ByteDance's Coze platform, focusing on the development and operation of AI agents. It provides developers with full lifecycle management from development, debugging, evaluation to monitoring...
Intern-S1 is a scientific multimodal large-scale model officially open-sourced and released by the Shanghai Artificial Intelligence Laboratory at the World Artificial Intelligence Conference. It integrates language and multimodal performance, possesses a high level of balanced development capability, and incorporates multidisciplinary expertise...
Hunyuan World 1.0, officially released and open-sourced by Tencent at the World Artificial Intelligence Conference, is the industry's first immersive, interactive, and simulated world generation model. The model integrates panoramic vision...
RiRiXin V6.5 is a new multimodal reasoning model launched by SenseTime. The model features an original image-text interwoven thought chain, in which images participate in reasoning in the form of ontologies, significantly improving cross-modal reasoning accuracy and surpassing Gemini 2.5 Pro.
GLM-4.5 is Zhipu's next-generation flagship model, specifically designed for agent applications. It is the first open-source state-of-the-art (SOTA) model to natively integrate inference, code, and agent capabilities. It adopts a hybrid expert (MoE) architecture and comes in two versions: ...
Wan2.2 is an advanced AI video generation model open-sourced by Alibaba. It includes three open-source versions: text-based video (Wan2.2-T2V-A14B), image-based video (Wan2.2-I2V-A14B), and unified video generation (Wan2.2-IT2V-...).
WebShaper is an innovative AI training data synthesis system launched by Alibaba Tongyi Labs. Through formal modeling and agent extension mechanisms, it provides high-quality, scalable data for training AI agents...
Skywork UniPic is an open-source, multimodal, unified pre-trained model from Kunlun Wanwei, possessing three core capabilities: image understanding, text-to-image generation, and image editing. The model is based on an autoregressive paradigm, integrating a MAR encoder and SigLIP...
Qwen3-30B-A3B-Instruct-2507 is an open-source Qwen3-30B-A3B non-thinking mode language model from Alibaba Tongyi. It has a total of 30.5 billion parameters, 3.3 billion activation parameters, a 48-layer structure, and a context length of 262,144.
SeedEdit 3.0 is an image editing model developed by ByteDance's Seed team. It supports fast, high-quality image editing using natural language commands. The model relies on the powerful text-to-image model Seedream 3.0, enabling precise processing...
Agent Lightning is a flexible and scalable intelligent agent optimization framework developed by Microsoft's research team. The framework can be seamlessly integrated into any existing agent framework (such as OpenAI Agents SDK, LangChain, etc.)...
HYPIR (Harnessing Diffusion-Yielded Score Priors for Image Restoration) is an advanced large-scale image restoration module developed by the research team of Professor Dong Chao at the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences...
Qwen3-30B-A3B-Thinking-2507 is an open-source inference model from Alibaba Tongyi, designed specifically for complex inference tasks. The model has 30.5 billion parameters, of which 3.3 billion are activated, supports a native context length of 256K, and can be expanded to 1M...
Abogen is a powerful text-to-speech tool that supports the rapid conversion of ePub, PDF, or text files into high-quality audio, and can generate synchronized subtitles. Based on the Kokoro-82M model, abogen supports multiple languages and speech styles...
FLUX.1 Krea [dev] is the latest text-to-image generation model launched by Black Forest Labs in collaboration with Krea AI. It supports the generation of more realistic and diverse images, achieving photorealistic levels.
Qwen3-Coder-Flash is a high-performance programming model developed by the Alibaba Tongyi Qianwen team. Its full name is Qwen3-Coder-30B-A3B-Instruct. The model possesses excellent agentic capabilities and excels in proxy programming, tool invocation, etc.
Seed Diffusion is an experimental diffusion language model developed by ByteDance's Seed team, focusing on code generation tasks. The model utilizes key technologies such as two-stage diffusion training, constrained sequential learning, and reinforcement-based efficient parallel decoding to achieve...
DispatchMail is an open-source, locally running AI email assistant that automates inbox management using AI technology. The tool monitors emails in real time and uses OpenAI's AI Agent to process emails based on user-defined prompts...
iFlow CLI is an AI agent developed by the Flow AI team that runs on the edge and is designed for developers and high-performing workers. iFlow CLI supports natural language interaction and can quickly analyze code, generate documentation, debug programs, manage...
Gemini 2.5 Deep Think is an AI model from Google designed to solve complex tasks. It's a variant of the model that won a gold medal at the 2025 International Mathematical Olympiad (IMO), utilizing Parallel Thinking technology...
InteriorGS is a high-quality 3D Gaussian semantic dataset launched by Quncore Technology. It contains 1000 3D Gaussian semantic scenes, covering more than 80 indoor environments, such as homes, convenience stores, wedding halls, and museums. The dataset contains 755...