Latest AI Resources

Ming-flash-omni-Preview is an open-source full-modal macromodel released by Ant Group inclusionAI, with a parameter scale of hundreds of billions, based on the sparse MoE architecture of Ling 2.0, with total parameters of 103B and activations of 9B. in full-modal understanding and generating...

ForgeTrain 是面壁智能联合清华大学、OpenBMB 开源社区发布的全球首个完全由 AI 编写、零人类代码介入的生产级大模型预训练框架。框架采用独创的 Forge Engineering（锻造...

Unlimited-OCR 是百度开源的端到端长文档 OCR 模型，采用 3B 参数 MoE 架构（每 token 仅激活约 500M 参数），核心创新在于将标准注意力替换为 R-SWA（参考滑动窗口...

SeedMusic 1.0是字节跳动推出的第一代AI音乐生成模型，专注于将用户的文字创意快速转化为带有人声的完整歌曲草稿。用户只需输入一段描述、歌词或风格方向，可在几分钟内生成包含主歌、副歌、编曲和人...

OpenAgents is the open source project that creates a network of AI agents and facilitates open collaboration between agents. A basic network infrastructure is provided to enable AI agents to seamlessly connect and collaborate. Users can quickly start their own agent network, extend functionality through a modular architecture, support...

UniWorld V2 is a new generation of image editing model jointly launched by RabbitZhan Intelligence and UniWorld team of Peking University. It has significant advantages in the field of image editing, especially in Chinese comprehension and execution of complex commands. The model can accurately render artistic Chinese fonts and support fine...

DeepSeek-Math-V2 is an open source mathematical reasoning model by DeepSeek, an AI company under Phantom Cube, and the latest version is based on DeepSeek-V3.2-Exp-Base improvement, with performance surpassing that of Gemini DeepThink to reach the international number...

MiniMax M2.1 is MiniMax's open source coding and agent model with 10 billion activations and support for many major programming languages such as Rust, Java, Golang, C++, Kotlin, Objective-C, TypeS...

Comprehensive Introduction Meiman is an online platform specializing in home design and marketing negotiation. The site provides a wealth of design materials, soft furnishings and proposal PPT templates, poster templates, etc. to help designers and homeowners quickly generate high-quality design solutions. Meiman's online soft furnishing design tool can be used in as little as 10 seconds...

html-video是Open Design团队开源的本地视频生成工具，能将HTML、CSS和数据直接转换为MP4视频。支持21种预设模板，涵盖产品宣传、知识讲解等场景，可配合Claude Code...

Lyria 3 Pro是谷歌推出的最先进的AI音乐生成模型。相比前代Lyria 3仅30秒的生成时长，Pro版本可一次性生成长达3分钟的完整音轨，精准控制前奏、主歌、副歌、桥段等歌曲结构。

FLUX.2 is an open source image generation and editing model released by Black Forest Labs that supports textual raw images, multi-image referencing, and image editing with richer details, clear textures, and stable lighting. There are four versions: FLUX.2 [pro] (comparable to the top closed source...

Fun-AudioGen-VD 是阿里通义实验室推出的创新语音大模型，专注于声音设计与场景化音频生成。模型支持通过自然语言指令直接生成包含特定音色、情绪表达和完整听觉场景的高质量音频，无需参考音频即可...

Open-o3 Video is an open source video inference model jointly developed by Peking University and ByteDance, focusing on enhancing video inference through temporal and spatial evidence. By explicitly labeling key evidence with timestamps and bounding boxes, it helps the model better understand and interpret video content.

Computer Use Preview is Google's open source AI browser automation tool based on the Gemini model , through natural language commands to achieve web page interaction . Using "screenshot→analysis→execution" visual recognition process , support Playwrigh...

ChronoEdit, an open-source AI image editing framework developed by NVIDIA in conjunction with the University of Toronto, redefines the image editing task as a video generation task to ensure that the editing results are temporally and physically consistent. By distilling a pre-trained video generation model with 14B parameters from a...

LongCat-Flash-Omni is an open source fully modal big language model released by the LongCat team of Meituan. With a parameter scale of 560 billion (27 billion activated parameters), it realizes millisecond-level real-time audio and video interaction capabilities while maintaining a large number of parameters.

Hy ASR 3.0 preview 是腾讯混元推出的新一代语音识别模型，深度融合最新大语言模型 Hy3 的语言理解能力与高精度语音识别技术，实现从"逐字转写"到"听懂语境"的跨越。模型采用 MoE ...

Lightpanda是Lightpanda.io公司开发的开源无头浏览器，专为AI自动化和Web抓取设计。采用Zig语言从零构建，去除了图形渲染等冗余功能，核心优势是速度比Chrome快11倍，内存占...

Qwen3.8-Flash-Next 是阿里 Qwen 发布的开源权重多模态 MoE 模型，定位为 Qwen4 架构的先导预览：主模型 125B 参数，另配 51B N-gram Embedding...

GLM-4.6V is a series of multimodal large language models open-sourced by Smart Spectrum AI. The series contains two versions: GLM-4.6V (106B-A12B), the basic version for cloud and high-performance cluster scenarios, with the Mixed Expert (MoE) architecture, a total of about 106 billion references, and an activation...

Grok 4.3是xAI公司推出的最新一代大语言模型，属于Grok系列的重要更新版本，支持100万token的上下文窗口，可一次性处理整本书、大型代码库或长文档，思维链推理强制开启，模型在给出答案前会...

Mistral 3 is the latest multimodal large model series released as open source by Mistral AI, including the flagship model Mistral Large 3 (675B total parameters) and the lighter version of the Ministral series (3B/8B/14B), both supporting image understanding...

MiniMax Music 2.5+是MiniMax推出的AI音乐生成模型，专注器乐创作。模型精通古典管弦、电子氛围、自然声景等多元风格，擅长将东方传统乐器与西方现代编曲融合，实现跨风格创新。

Kimi K3 是月之暗面（Moonshot AI）发布的全球首个开源 3 万亿级大模型，总参数达 2.8 万亿。采用 KDA 混合线性注意力机制与 Stable LatentMoE 稀疏架构，896...

Fun-CosyVoice3.5是阿里通义实验室推出的第三代语音合成大模型，主打"自然语言指令控制"能力，用户可直接用口语化描述（如"语气坚定一点""语速慢一点"）来调节合成语音的情绪、语速和风格，无...

Kosong is a new AI Agent development framework open-sourced by Dark Side of the Moon (Moonshot AI) that provides developers with a lightweight, flexible, and highly scalable underlying support for building next-generation intelligent body applications. With an asynchronous tool orchestration engine that efficiently schedules multiple tools...

DeepSeek Harness（DSH）是 DeepSeek AI 开源的智能体运行框架，当前为开发者预览版并采用 MIT 许可，核心理念是“一切皆插件”：模型、工具、技能、会话、沙箱、存储、循环...

Seedream 5.0 Pro是字节跳动豆包大模型团队发布的多模态图像创作模型，定位为面向专业创作者和企业级用户的设计工具。相比前代，在图文匹配、结构合理性等基础能力上全面提升

Wan3.0 是阿里云通义全网公测的新一代视频生成大模型，被称为"阿里最强视频模型"。在生成时长、万能创作、全能参考、真实感四大维度全面升级。单次可稳定生成最长30秒的一镜到底视频

Qwen-Audio-3.0-ASR-Flash 是阿里巴巴通义千问团队发布的语音识别大模型，主打"长音频不丢词、行业词不用教"。模型在上下文一致性、行业词识别与热词定制化三大维度实现系统性升级，集成...

KAT-Dev-72B-Exp is an open-source programming-specific large language model launched by the Racer team, optimized based on reinforcement learning technology, which achieved an accuracy rate of 74.6% in the SWE-Bench Verified benchmark test, the best performance of any open-source model at present. The model uses innovative...

BestBlogs is a platform focused on providing high-quality content for technology practitioners, entrepreneurs, product managers, and more. Through RSS feeds and crawler technology, it collects articles, podcasts, videos, and other multi-format content from more than 400 high-quality blogs. The core strength lies in utilizing AI big language...

VibeThinker-1.5B is a 1.5 billion parameter large-scale language model open-sourced by Weibo AI. Fine-tuned based on Alibaba's Qwen2.5-Math-1.5B, it is optimized for mathematical and coding tasks and performs well, with industry-leading inference performance.

PartCrafter is an advanced 3D generative model, jointly proposed by Peking University, ByteDance and Carnegie Mellon University. It can generate multiple semantically explicit and geometrically diverse 3D mesh parts from a single RGB image at once. The models are modeled through a combinatorial potential space and...

Dexbotic is the open source Visual-Linguistic-Action (VLA) model of embodied intelligence one-stop scientific research service platform of Dexmal, which solves the problems of fragmentation and low efficiency of research in the field of embodied intelligence. Based on PyTorch, Dexbotic is a one-stop research service platform to solve the problems of fragmentation and inefficiency in the field of embodied intelligence...

MemMachine is an open source AI memory system developed by MemVerge, designed for AI models and intelligences, which can store and recall interaction data like the human brain, solving the problem of AI "stateless memory loss". It adopts a layered architecture (short-term memory, long-term memory, user image...

Jamba Reasoning 3B is a lightweight inference model open-sourced by Israeli AI startup AI21 Labs with strong performance and potential for a wide range of applications. It utilizes a hybrid SSM-Transformer architecture that combines Trans...

商汤输入法AudioClaw是商汤科技基于日日新多模态大模型打造的AI语音智能应用，深度接入OpenClaw底层能力，定位为"会听话的龙虾"智能助手。

DeepOCR is an open source replication project that implements the core architecture of DeepSeek-OCR, which efficiently processes textual information through optical compression techniques. The core is DeepEncoder, consisting of SAM-base (processing high-resolution images), 16× convolutional compressor...

GigaWorld-0 is the open source world modeling framework of domestic Embodied Intelligence startup GigaAI, mainly used to solve the data bottleneck problem in the field of Embodied Intelligence (Embodied AI). Efficiently generating high-quality, diverse and physically realistic training data, push...

Koina is an open source decentralized machine learning platform focused on simplifying proteomics data analysis. Developed by a team from the Technical University of Munich, Germany and the University of Michigan, USA. The platform integrates more than 30 mainstream models (e.g., ProSIT, MS²PIP) through a standardized interface and supports peptide mass...

LOGOS（Language Of Generative Objects in Science）是阿里集团联合中国人民大学高瓴人工智能学院开发的首个基于统一"科学语法"的多领域科学生成基础模型。

Gemini 3.1 Flash Live 是 Google 推出的旗舰级实时语音模型，被誉为"迄今最高质量的音频和语音模型"。主打超低延迟的自然对话体验，支持 90+ 种语言的原生多模态交互，能精准...

Nemotron 3 Super是NVIDIA 发布的Nemotron 3 系列中目前最强大的开源权重 AI 模型。模型采用 1200 亿参数的混合专家（MoE）架构，推理时仅激活 120 亿参数，实...

跃迁维度（dimleap） 是AI模型聚合服务平台，为全球开发者提供高效能、低成本的多品类AI模型API服务。平台目前已接入DeepSeek、GLM（智谱清言）、MiniMax、kimi、qwen（通...

QwenLong-L1.5 is an open source long text inference model from Alibaba Tongyi Lab, focusing on solving complex inference problems with ultra-long contexts (e.g., 1M-4M tokens). The core breakthrough lies in three major innovations in the post-training phase: through knowledge graph, SQL parsing and multi-intelligence...

TurboQuant 是 Google Research 推出的突破性内存压缩算法，专为解决大语言模型推理中的 KV 缓存瓶颈而设计。

InkSight is Google's open source AI handwriting recognition tool that converts paper handwritten notes into editable digital inked files (e.g. SVG format). Unlike traditional OCR , can recognize text content , can restore the handwriting style , paragraph structure and focus marking , support for multi-language processing .

Wujie-Emu3.5 is an open source multimodal world grand model from Beijing Zhiyuan Artificial Intelligence Research Institute, with 34 billion references and native world modeling capability. Trained by 10 trillion multimodal Token (including 790 years of video data), it can simulate the laws of physics and realize graphic generation, visual guidance...

GLM-5.3-Flash是智谱Z.ai上线并开源的旗舰级高性价比模型，也是GLM-5系列首个原生多模态模型；采用320B总参数、18B激活的MoE，支持1M上下文与128K输出，通过稀疏+线性注意力...

SAIL-VL2 is an open source multimodal visual language model by the Byte Jump team, focusing on joint modeling of multimodal inputs such as images and text. Using the sparse mixture of experts (MoE) architecture and progressive training strategy, it achieves high performance at parameter scales from 2B to 8B, especially in the areas of graphic comprehension, math...

GPT-5.4 mini 是 OpenAI 发布的轻量级模型，定位为"迄今能力最强的小型模型"。在保留 GPT-5.4 核心能力的同时，实现了速度提升 2 倍以上，成本大幅降低，专为对延迟敏感的高频工...

ChatExcel Ultra 是ChatExcel团队面向企业用户和专业数据办公场景推出的全新桌面版，定位为国内首款企业级AI数据智能体。

FG-CLIP 2 is the world's leading graphical cross-modal visual language model (VL-M) launched by 360 Artificial Intelligence Research Institute, which surpasses similar models from Google and Meta in 29 authoritative benchmark tests, making it the most powerful VL-M at present.It is able to accurately recognize the gross...

Qwen3.8-Max-Preview是阿里巴巴通义千问团队发布的新一代旗舰大模型预览版 。模型总参数达2.4万亿，采用稀疏MoE架构，是通义千问系列首个突破万亿参数的多模态模型，支持文本、图片、视频...

GPT-5.3 Instant 是 OpenAI 发布的 ChatGPT 主力模型升级版，主打"体验优化"而非参数堆叠。精准解决了用户长期诟病的"说教感"和机械回复问题，对话语气更自然直接，大幅减少了...

Flowra is ModelScope joint woo mile WULI team open source graph execution engine and node package development tools, is the core component of FlowBench. Through the directed acyclic graph (DAG) organization workflow , with intelligent caching , parallel scheduling , distributed support ...

HyOCR-1.5是腾讯混元团队开源的轻量化端到端OCR专家大模型，参数量仅1B，是领域首个训练、推理、权重完整开源的专家模型 。采用端到端架构，输入图片即可直接输出Markdown正文、HTML表格...

VibeVoice-Realtime is Microsoft's open source lightweight real-time text-to-speech (TTS) model designed for low latency and real-time interaction. Supports streaming text input , from the first text token can be vocalized , the delay is only about 300 milliseconds , suitable for dynamic number ...

SeedRealtime 是字节跳动 Seed 团队发布的原生音视频全双工大模型，首发落地于豆包 App 的「打电话」视频通话功能。与传统"语音识别→视觉理解→大模型推理→语音合成"的级联流水线不同...

Seekdb (OceanBase Seekdb) is Ant OceanBase open source AI native hybrid search database, supporting unified hybrid search of vector, full text, scalar and geospatial data, using multi-stage retrieval mechanism, to achieve high precision search under low latency ...

Agnes 2.5 Pro Alpha 是新加坡 AI 实验室 Agnes AI 发布的高性价比推理模型，支持文本、图像输入与文本输出，提供百万 Token 上下文和 65,536 Token 最大输...

MiniMax H3是稀宇科技MiniMax发布的通用全模态生成模型，可在统一上下文中理解文本、图像、视频与声音，生成最高15秒、2K分辨率、带原生双声道音频的音视频。

Qwen 3.8-Max 是阿里巴巴通义千问发布的旗舰级大模型，也是 Qwen 系列迄今规模最大、能力最强的模型。采用稀疏混合专家（MoE）架构，总参数达2.4万亿，推理时仅激活约950亿参数，在保持...

TDream是腾讯内测的AI内容创作平台，以"互动影游化创作"为核心定位，试图重构用户从内容生产到交互体验的全链路。区别于市面上仅输出单一线性视频的传统文生视频工具，TDream依托五大AI引擎，支持...

Claude Sonnet 5是Anthropic 发布的最具 Agent 能力的 Sonnet 模型，以中端价格提供接近旗舰 Opus 4.8 的自主智能体能力。模型支持 100 万 token 上...

GPT-Live 是 OpenAI 推出的新一代语音模型，全面升级 ChatGPT 的语音交互体验。基于全双工（full-duplex）架构构建，能同时倾听和说话，彻底打破了传统 AI 语音助手"你说...

Claude Opus 5 是 Anthropic 推出的新一代旗舰大模型，核心是用旗舰一半的价格获得接近顶尖水平的智能"。模型在维持与前代 Opus 4.8 相同调用成本的同时，实现性能的翻倍跃升...

Muse Image 是 Meta 推出的首个自研 AI 图像生成模型，由 Meta Superintelligence Labs 开发。采用独特的智能体（Agentic）架构，在生成图像前会先与 M...

Voxtral TTSoxtral TTS是法国AI公司Mistral AI发布的开源文本转语音模型，采用40亿参数轻量化架构，量化后仅需3GB内存即可在智能手机等边缘设备实时运行。模型原生支持英语...

Grok 4.5 是 SpaceXAI（原 xAI）发布的旗舰大语言模型，定位为"Opus 级别"的编码与智能代理专用模型。模型基于 1.5 万亿参数的 V9 架构打造，采用 MoE（混合专家）架构...

Grok Bot 是 SpaceXAI（原 xAI）发布的 AI Agent 产品，定位为「永远在线的 AI 同事」。不同于传统聊天机器人，运行在独立的云端 Linux 虚拟机中，配备浏览器、文件系统...

SciClaw（科研龙虾） 是面向科研人员的云端 AI 协同系统，定位为"科研人的深夜搭子"。能深度分析文献、自主执行实验、自动排版生成论文/PPT/海报，记住用户电脑里的所有研究资料，将灵感、执行与...

Qwen-Audio-3.0-TTS是阿里巴巴通义千问团队发布的旗舰级语音合成大模型，提供面向实时交互的Flash版（首包延时约300ms）与面向高质量生成的Plus版，Plus版登顶全球权威榜单Ar...

OnSolo是腾讯推出的AI原生一站式影视创作平台，前身WorkSolo，以"a Crew of One"为核心理念，让个人仅凭一句话灵感即可完成从剧本、角色资产、分镜到视频成片的全部创作流程。

LongCat-Image is an open source image generation and editing model released by the LongCat team of Meituan. Using a hybrid backbone architecture (MM-DiT+Single-DiT), combined with a visual language model (VLM) conditional encoder, it is able to realize text-generated images and multiple rounds of image editing...

Gemini 3.5 Flash-Lite 是 Google 发布的高吞吐低成本模型，官方定位为 3.5 系列中最快、最具成本效益的型号，面向 agentic search、文档处理与大规模生产流量...

Chatterfly 是腾讯推出的桌面端 AI 语音助手，支持 Windows 和 macOS。用户只需按住 Fn 键说话，工具能把口语化的表达自动整理、润色并排版成邮件、汇报、纪要等规范文本。

XiaoMi MiMo Desktop 是小米推出的桌面级通用 AI 助手，搭载自研的 MiMo-X-Pro 与 MiMo-X-Flash 双模型。用户无需提前整理素材格式，可直接把表格、PDF、音视...

JellyToken 是国内领先的大模型 API 聚合与分发平台，定位为"国内主流 AI 大模型一站式超市"。用户仅需一个 API Key 即可无缝调用通义千问、DeepSeek、智谱、月之暗面、豆包...

InstanceAssemble is a layout control generation technology jointly open-sourced by Little Red Book and Fudan University, which realizes accurate image generation from simple to complex and from sparse to dense layout through the mechanism of "Instance Assemble Attention". Adopting a two-stage cascade architecture, Mr. Mr. into the image background, and then one by one ...

Hyra（Hunyuan Research Agent）是腾讯混元团队发布的科研智能体，首个版本为Hyra-1.0，支持递归自我改进（RSI），专为性能导向的研究与工程任务打造 。采用轻量通用框架，通...

Qwen-Image-3.0 是阿里千问发布的第三代图像生成基础模型，核心主线为"实"，内容丰实、细节真实、知识厚实。模型支持最大 4.5k token 超长文本输入，较前代提升 4.5 倍，可一次性...

Astryx 是 Meta 以 MIT 协议开源的 React 设计系统，历经八年内部打磨，长期支撑 Facebook、Instagram、WhatsApp 和 Threads 等超过 13,000 ...

Nano Banana 2 Lite（gemini-3.1-flash-lite-image）是Google发布的AI图像生成模型，定位为Nano Banana家族中速度最快、成本最低的轻量版。可在约...

Gemini 3.6 Flash 是 Google 发布的新一代主力模型（Workhorse Model），是 Gemini 3.5 Flash 的升级版，主要面向生产级 AI 智能体（Agent）和...

Ling-3.0-tiny 是蚂蚁 InclusionAI 百灵 Ling 3.0 系列中最小的 MoE 模型，约 7.9B 总参数、每 token 激活约 1.3B，强调低延迟、低成本与 Agent...

MiniMax Design 是 MiniMax（稀宇科技）发布的多模态创作 Agent 工作台，也是开源视频模型 H3 的首个官方工具框架（Harness）。用户只需用自然语言提出创作需求，Agen...

Step Code 是阶跃星辰推出的开源终端编程智能体。Step Code能直接在命令行里帮你写代码、调试、跑任务，直到交付成果，不用来回切换开发工具。Step Code在 Terminal Benc...

MAGI-2 Preview 是 Sand.ai 开源的统一音视频生成模型，总参数约 114B、单 token 仅激活约 6B，是全球首个千亿级开源 MoE 视频模型 。采用单流架构将文本、视频、音频...

FLUX 3 是德国 AI 公司 Black Forest Labs 发布的首款多模态基础模型，基于自研 Self-Flow 架构，在单一网络中统一训练图像、视频、音频与动作预测，实现真正的"全能模型...

Ovis-Image is a 7 billion parameter text-generated graph model open-sourced by the AIDC-AI team of Alibaba International Digital Commerce Group, focusing on high-quality text rendering. Based on Ovis-U1 architecture, it inherits the advanced visual decoder and bi-directional Token refiner ...

GLM-5.3 是智谱发布的新一代旗舰模型，沿用 GLM-5.2 约7430亿参数基座，能力跃迁主要来自后训练 Scaling：通过 IndexShare、SAO、Slime 框架与更多长程任务环境强...

Lovart中文版是基于多模型架构的AI设计Agent，用户只需像聊天一样描述创意需求，平台可自动生成海报、视频、Banner等高品质视觉内容。平台深度整合Seedance、MiniMax H3等前沿...

Wan-Streamer v0.2 是阿里通义实验室发布的端到端全模态实时交互模型，首次将「听、看、说、演」统一进单个因果 Transformer 架构，实现类人般的实时双工交互。

Qwen3.8-LiveTranslate 是阿里通义千问推出的实时同声传译大模型。模型采用Interleave单流架构，将字均延迟从2.8秒压缩至2.3秒，支持60种语言识别、29种语言语音输出。

dots.tts 是小红书 dots 团队（rednote-hilab）与上海交大 X-LANCE 实验室联合研发的 20 亿参数全连续端到端自回归语音合成基础模型，架构为「语义编码器 + Qwen2...

Play with Putty 是 Google Labs 推出的协作式 vibe coding 实验工具，口号是"让你的想象力进入多人模式"。允许多人通过共享链接进入同一空间实时协作，任何人都能用自...