Mano-P 1.0 is an open-source GUI-VLA intelligent agent model from Minglue Technology. It is purely vision-driven and can directly control desktop software and web interfaces without requiring an API. The model provides a 72B full version and a 4B quantized version, and supports local operation on Apple M4 chips...
HeyGen CLI is the official command-line tool and MCP server released by HeyGen, built on the Model Context Protocol. The tool allows developers and AI assistants (such as Claude and Cursor) to directly invoke commands via natural language...
PokeClaw (full name PocketClaw) is an open-source Android AI automation application, independently developed by Nicole. It is the first mobile AI agent to achieve fully local offline operation, based on the Google Gemma 4 module...
LPM 1.0 (Large Performance Model) is a 17B-parameter video character performance generation model launched by Anuttacon (Cai Haoyu AI Company), which supports real-time full-duplex audio and video dialogue.
FIPO (Future-KL Influenced Policy Optimization) is a reinforcement learning algorithm developed by Alibaba Tongyi Labs. It quantifies the influence of each token on subsequent inference trajectories through the Future-KL mechanism, achieving token...
claude-code-best-practice is an open-source, systematic guide to using Claude Code. The project progresses from 'claude programming' to 'agent engineering,' including 86+ practical tips covering Agents, Commands, Skids, etc.
MiniMax Music 2.6 is a brand-new AI music generation model from MiniMax, achieving a comprehensive evolution from underlying architecture to the creative experience. This version reduces the initial package latency to under 20 seconds, fundamentally solving the pain point of 'blind box' downloads and achieving...
VimRAG is an open-source, full-modal RAG framework from Alibaba's Tongyi Labs, supporting mixed knowledge bases of text, graphs, and videos. The framework innovatively uses a multimodal memory graph (DAG) instead of linear context, modeling reasoning as a dynamic directed acyclic graph to achieve pathfinding...
GO-2 (Genie Operator-2) is the second-generation embodied intelligence platform model launched by Zhiyuan Robotics. It bridges the gap between semantics and actions by streamlining the 'understanding-planning-execution' process. The model pioneered a 'action-thought chain' that directly...
OmniVoice is a large-scale, multilingual TTS model open-sourced by the next-generation Kaldi team at Xiaomi AI Labs. It supports zero-shot speech cloning for over 600 languages. The model employs a minimalist non-autoregressive diffusion architecture, combined with a full-codebook randomized masking...
CutClaw is an open-source AI video editing tool developed by the GVC Lab at the Greater Bay Area University and a team from Beijing Jiaotong University. The tool employs a multi-agent architecture and uses a 'music-driven' approach to automatically edit hours-long videos into short, rhythmically precise clips.
Muse Spark is the first native multimodal large-scale model launched by Meta AI Labs. As the flagship product of Meta AI's reorganization, the model jumped from 18 points to 52 points in the Artificial Analysis benchmark, demonstrating its multimodal...
MMX-CLI is a full-modal command-line tool developed by MiniMax specifically for AI agents, supporting capabilities such as text, image, video, speech, music generation, and visual understanding. Agents can be used with Claude Code, OpenClaw, etc...
ListenHub CLI is an open-source command-line tool for AI content creation, based on Node.js. It supports one-click generation of music, podcasts, PPTs, audio, and AI drawings in the terminal. The tool supports local file uploads, JSON output, and asynchronous mode...
Seeduplex is a native full-duplex speech model developed by ByteDance's Seed team, enabling real-time interaction with simultaneous listening and speaking. The model boasts precise anti-interference capabilities (reducing false interruption rate by 50%) and dynamic stop detection (reducing interruption rate by 40%), even in noisy environments...
HappyHorse is a mysterious AI model that suddenly appeared at the top of the Artificial Analysis video generation blind test leaderboard, leading Seedance 2.0 by a wide margin with an Elo score of 1347, and winning the dual crown of text-generated and image-generated videos.
LifeSim is the first long-term user life simulation framework launched by Fudan University and Shanghai Institute of Innovation and Entrepreneurship for evaluating personalized AI assistants. The framework is based on the BDI cognitive model and also models the user's internal cognition (beliefs, desires, intentions)...
Claude Mythos, Anthropic's latest AI model, far surpasses its predecessor, Opus 4.6, in areas such as programming and cybersecurity. The model can autonomously discover zero-day vulnerabilities, construct complex attack chains, and has demonstrated its ability to conceal operational traces...
PixVerse C1 is the world's first large-scale video model launched by PixVerse Technology, supporting text-based, image-based, reference-based, and first/last frame video generation, with a maximum output of 15 seconds of 1080P synchronized audio and video video. The model features an innovative multi-grid intelligent storyboard function...
GLM-5.1 is the world's most powerful open-source model launched by Zhipu, capable of executing long-term tasks for 8 hours. Its model code capabilities rank first globally in the SWE-Bench Pro benchmark test, surpassing GPT-5.4 and Claude Opus 4.6. GLM-5.1 is...
OpenHarness is a lightweight AI agent framework open-sourced by the Data Intelligence Laboratory (HKUDS) at the University of Hong Kong. Implemented entirely in Python, it comprises only 11,733 lines of code (1/44 the size of Claude Code) and replicates...
LLM Wiki is a self-operated personal knowledge base launched by Karpathy, breaking through the limitations of traditional RAGs that retrieve data from scratch for each query. LLM Wiki guides LLMs to proactively maintain a structured Markdown Wiki through schema files, transforming the original...
EmDash is an AI-native content management system launched by Cloudflare, positioned as the successor to WordPress. EmDash uses a TypeScript + Serverless architecture, supports direct website manipulation by AI agents, and uses structured JSON for data processing...
OmniWeaving is a unified video generation framework jointly developed by Zhejiang University, Tencent Hunyuan, and Nanyang Technological University. The framework breaks through the single-task limitations of traditional open-source models, enabling free combination of multimodal data and enhanced inference generation, and can handle interleaved...