LTX-2 is an advanced AI video generation model developed by Lightricks, designed specifically for high-quality video creation. It can generate cinematic-quality videos at native 4K resolution and 50fps, and supports multimodal input, including text, ...
Game-TARS is a general-purpose game intelligence agent developed by ByteDance's Seed team. Based on a unified keyboard-mouse action space training model, it can be pre-trained on a large scale in operating systems, web pages, and simulation environments. It leverages over 500 billion...
SWE-1.5 is a high-performance AI programming model designed specifically for software engineering, launched by AI unicorn Cognition. It boasts tens of billions of parameters, near-state-of-the-art coding capabilities, and significant speed breakthroughs...
LongCat-Flash-Omni is an open-source, multimodal language model from Meituan's LongCat team. Based on the efficient architecture of the LongCat-Flash series, it innovatively integrates multimodal perception and speech reconstruction modules, boasting 5600...
AI Research Foundations is a free online course jointly offered by Google DeepMind and University College London (UCL). The course helps learners gain a deep understanding of Transformer models and master modern language skills through practical exercises...
Open AgentKit (OAK) is an open-source AI agent development platform from ZeroOne, providing developers with an open and universal agent development solution. Inspired by OpenAI AgentKit, the platform supports various models and ecosystems...
BettaFish is an open-source, multi-agent public opinion analysis system that helps users quickly understand public sentiment towards events, brands, or products. The system is driven by natural language; users simply submit their requests as if in a chat...
Ouro is a looped language model (LoopLM) released by ByteDance's Seed team in collaboration with several other institutions. The name originates from the "Ouroboros," a symbol of looping and self-devouring. The Ouro model...
SAIL-Embedding is a multimodal embedding model jointly developed by ByteDance's Douyin SAIL team and the MMLab of the Chinese University of Hong Kong. It addresses practical application problems in multimodal information retrieval and recommendation systems...
FG-CLIP 2 is an open-source, fine-grained bilingual visual-language alignment model launched by 360, designed specifically to solve the problem of accurate alignment between vision and language. It has achieved significant breakthroughs in the field of visual-language understanding, especially in Chinese-English bilingual tasks...
NavFoM (Navigation Foundation Model) is the world's first cross-ontology, full-domain surround-view navigation foundation model, jointly released by Galaxy General and teams from Peking University, the University of Adelaide, Zhejiang University, and others. It features full-scene support...
Handy is an open-source speech-to-text desktop application that runs completely offline and supports Windows, macOS, and Linux systems. Built with Rust and React/TypeScript, it features a clean interface and easy operation. Users can quickly...
UniWorld V2 is a next-generation image editing model developed by RabbitShow Intelligence and the UniWorld team at Peking University. It employs the innovative UniWorld-R1 training framework and is the first to apply reinforcement learning policy optimization to image editing, through Diffu...
iFlytek Spark X1.5 is a large-scale deep inference model launched by iFlytek, based on a fully domestically produced computing platform. The model is the first to overcome the end-to-end training efficiency problem of MoE models, achieving end-to-end performance exceeding 93% of international competitors. ...
UNO-Bench is a full-modal large model evaluation benchmark launched by Meituan's LongCat team. Addressing the shortcomings of existing evaluation systems, UNO-Bench uses high-quality, diverse data to accurately measure the single-modal and full-modal performance of models...
Kosmos is a next-generation AI scientist, an upgrade from Robin, an automated research system launched by FutureHouse. Kosmos employs a structured world model, enabling it to efficiently integrate massive amounts of information; a single run can parse 1500 documents...
Kimi-k2 Thinking is an AI model developed by Dark Side of the Moon, possessing general agentic capabilities and deep reasoning abilities. The model can perform multi-turn tool calls, supports context processing up to 256k bytes, and is suitable for complex tasks...
Open-o3 Video is an open-source video inference model jointly developed by Peking University and ByteDance. It achieves accurate video inference by integrating explicit spatiotemporal evidence (key timestamps and bounding boxes). Through a carefully designed STGR...
OmniVinci is a multimodal language model from NVIDIA, specifically designed for handling multimodal tasks involving vision, hearing, language, and reasoning. It achieves cross-modal semantic alignment through the unique OmnialignNet technology, Temporal Embedding, and more...
Step-Audio-EditX is the world's first LLM-level audio editing model, open-sourced by StepStar. It features fine-grained iterative control across three axes: emotion, speaking style, and paralinguistics, allowing for arbitrary control over the intensity of emotions such as anger, happiness, and sadness...
Maya1 is an open-source AI speech synthesis model from the Maya Research team, designed specifically to generate emotionally expressive speech. The model uses natural language descriptions to design voices and supports over 20 emotional expressions, such as laughter, crying, etc.
AI-Trader is an open-source AI trading framework developed by the University of Hong Kong. It allows different AI models to trade autonomously in real-world market environments, exploring the investment capabilities of AI. The framework supports multi-model competition and integrates market data and news...
LocalAI is an open-source local AI inference framework that provides complete localized AI solutions, supports multimodal models (such as text generation, image generation, and speech processing), and is compatible with the OpenAI API. It emphasizes privacy protection...
SuitAgent is an AI-powered legal litigation agent based on the Claude Code architecture. It comprises 10 specialized AI agents that collaborate to break down complex litigation cases into multiple independently executable workflows, enabling the efficient generation of legal documents...