Anthropic Skills is Anthropic's skill system for Claude. Essentially, it's a folder containing instructions, scripts, and resources that allow Claude to dynamically load and improve performance for specific tasks. Each Skill...
Kami is an open-source AI document design system from tw93, distributed as a Claude Code Skill, specifically designed to solve the problems of monotonous layouts and cheap visuals in AI-generated documents. The tool uses a strict set of print constraints...
Uni-1.1 is a new generation image generation model and API service launched by Luma AI. It adopts a decoder-only autoregressive Transformer architecture, which integrates text inference and pixel generation into a unified process.
DeepSeek-TUI is a fully terminal-based programming agent built by Hayden Brown using Rust, specifically designed for native DeepSeek V4 series models (deepseek-v4-pro / deepseek-v4-flash)...
GPT-5.5 Instant is the next-generation default model of ChatGPT released by OpenAI, replacing GPT-5.3 Instant, and is freely available to all users. The model combines the basic intelligence and rapid response of GPT-5.5, and reduces hallucinations in high-risk domains by 52%...
OpenWhip is an open-source desktop utility based on Electron, designed specifically for Claude Code users. When an AI gets stuck in an infinite loop, repeatedly modifies the same line of code, or becomes unresponsive for an extended period, the user can click the whip icon in the system tray...
Cube Sandbox is an open-source AI Agent sandbox execution environment platform from Tencent Cloud, built on RustVMM and KVM, offering both hardware-level security isolation and sub-millisecond startup speeds. Each Agent runs in an independent kernel...
OfficeCLI is the world's first open-source command-line Office suite designed specifically for AI agents, launched by the iOfficeAI team. It allows any AI agent to fully control Word, Excel, and PowerPoint with just one line of code...
Career-Ops is an open-source, AI-driven job search system built on Claude Code. It's not a traditional "mass application tool," but rather an intelligent filter that uses structured assessments to help job seekers identify truly valuable candidates from a vast pool of positions...
Flipbook is an AI-powered, infinitely visual browser developed by former OpenAI engineer Zain Shah and a team of YC alumni. All pages are generated entirely in real-time, on-demand. Flipbook completely abandons HTML, CSS, and JavaScript...
gnhf (short for Good Night, Have Fun) is an open-source AI Agent auto-orchestrator. Its core concept is to allow developers to set a goal for the AI before going to sleep, allowing it to iterate and work continuously while you rest. It does only one task at a time...
LongCat-2.0-Preview is a preview version of a trillion-parameter-level large model launched by Meituan. It adopts the MoE architecture, with a total of 1.6T parameters and about 48B activation parameters, and supports a 1M ultra-long context.
Vibe-Trading is an open-source AI-driven multi-agent financial workspace from the Data Science Lab (HKUDS) at the University of Hong Kong, translating natural language instructions into executable trading strategies, research insights, and portfolio analysis. The project covers A...
Qwen-Scope is an open-source toolkit for interpretable large models from the Alibaba Tongyi Qianwen team. Based on sparse autoencoder (SAE) technology, it extracts interpretable features from the hidden layers of Qwen3/Qwen3.5 series models.
MindDR 1.5 is a multi-agent deep research framework launched by the Li Auto Information Agent Team. With only about 30B parameters, it achieved a score of 52.54 on DeepResearch Bench, reaching the industry-leading level.
Hy-MT1.5-1.8B-1.25bit is an ultra-quantized and compressed offline translation model for mobile devices, developed by Tencent's Hunyuan team. The model is based on the Hunyuan translation model Hy-MT1.5-1.8B (1.8 billion parameters) and utilizes innovative Shear...
FlashQLA is a high-performance linear attention operator library based on TileLang, open-sourced by Tongyi Labs. FlashQLA achieves high performance through operator fusion, gate-driven in-card sequence parallelism, and Warp-Specialized optimizations in Hop...
Ling-2.6-1T is InclusionAI's latest open-source flagship model with trillions of parameters, designed specifically for agents, coding, and complex workflows. The model employs a hybrid framework of MLA and Linear Attention...
One-Eval is an open-source automated agent-driven large model evaluation framework developed by the OpenDCAI team at Peking University. It is built on DataFlow and LangGraph and focuses on NL2Eval (Natural Language to Evaluation) capabilities.
Step Image Edit 2 is a new generation of lightweight image generation and editing model launched by StepStar. With only 3.5B parameters, it outperforms large open-source models of 12B-20B in practical applications. The model generates an image in 0.5-2 seconds per pass...
Spark X2-Flash is a large language model based on the MoE architecture released by iFlytek. It has a total of 30B parameters, supports 256K ultra-long context, and is trained on a domestic computing power cluster based on Huawei Ascend 910B.
SenseNova U1 is a native unified multimodal model launched by SenseTime based on the NEO-Unify architecture, achieving understanding, reasoning, and generation within a single architecture. The model abandons traditional visual encoders and VAEs, constructing a unified representation space...
The Nemotron 3 Nano Omni is part of the NVIDIA Nemotron 3 model family, an open-source, high-efficiency multimodal inference model designed specifically for agent systems. The model employs a 30B-A3B hybrid MoE architecture, integrating visual, audio, and text...
Agent Skills is an open-source AI programming agent skillset developed by Addy Osmani, head of Google's Gemini team. Agent Skills encapsulates the workflows and engineering standards of senior Google engineers into 20 skills...