Reasonix is an open-source terminal programming agent specifically designed for DeepSeek. It utilizes an append-only runtime loop built around a byte-stable prefix-cache mechanism, achieving a cache hit rate of 99.82% and significantly reducing input costs for long sessions...
DataBuddy is a native AI Agent workbench launched by Tencent Cloud for enterprise data infrastructure. It is built on the same Harness framework as WorkBuddy and covers three core scenarios: data analysis, governance and engineering.
BitCPM-CANN is a 1.58-bit ternary large model jointly launched by Wallfacer Intelligence, Tsinghua University, and the OpenBMB open source community. It is China's first 1.58-bit ternary large model that has been trained end-to-end entirely on Huawei's Ascend domestic computing platform.
FreeLLMAPI is an open-source, OpenAI-compatible proxy service that aggregates free quotas from approximately 14 AI service providers into a single API endpoint, providing a total of approximately 1.3 billion tokens of free inference quotas per month.
Command A+ is an open-source multimodal large model from Cohere Labs. It employs a hybrid expert architecture, has a total of 218 bytes of parameters but only 25 bytes of activation parameters, balancing powerful performance with inference efficiency. The model supports 128K ultra-long sequences...
Software Copyright Materials Skill is an open-source software copyright skill that automatically generates source code documentation, user manuals, and application forms that meet Chinese software copyright application requirements by reading the source code of a real local project.
Horizon is an open-source, AI-driven information aggregation system that helps users build their own AI news radar. The tool supports crawling information from multiple sources, including RSS, Hacker News, GitHub, Reddit, Telegram, and Twitter/X...
Mega-ASR is a robust, full-scenario speech recognition platform model jointly open-sourced by Nanyang Technological University (NTU), National University of Singapore (NUS), and Shanghai Artificial Intelligence Laboratory. The model uses Qwen3-ASR 1.7B as its underlying architecture and is designed for...
CodeGraph is a pre-indexed code knowledge graph tool for AI coding agents, deeply integrated with Claude Code, Cursor, Codex CLI, OpenCode, and Hermes Agent via the MCP protocol. It utilizes tree-...
SenseNova-Skills is an open-source, modular AI office skills library developed by the OpenSenseNova team, designed specifically for Agent runtime. The tools integrate infographic generation, PPT creation, Excel data analysis, and deep learning...
LLM Council is an open-source multi-model collaboration framework by Andrej Karpathy. It uses OpenRouter to distribute user questions to multiple large models (such as Claude, GPT, Gemini, etc.) for independent responses, with each model remaining anonymous...
ReadFrog is an open-source AI-powered immersive translation browser extension. It can translate and interpret any content on a browser, supporting Chrome, Edge, and Firefox. It utilizes context-aware translation and word highlighting analysis...
Hy-MT2 is a new generation of large-scale translation model open sourced by Tencent Hunyuan. It offers three sizes: 1.8B, 7B, and 30B-A3B, and supports mutual translation between 33 languages and translation between 5 ethnic minority languages/dialects.
Stable Audio 3 is a new generation of open-source audio generation models from Stability AI. Based on a stream-matching latent space diffusion architecture, it supports text-to-music and sound effects conversion, audio editing, and continuation. Stable Audio 3 provides Sma...
Autoresearch is an open-source AI autonomous research experiment framework by Andrej Karpathy. The framework allows an AI agent to automatically run nanochat training experiments on a single GPU: autonomously tuning hyperparameters, executing training, analyzing and validating results...
LongCat-Video-Avatar 1.5 is an open-source audio-driven digital human video generation framework from Meituan's LongCat platform. The model is built upon the 13.6B-parameter base video generation model LongCat-Video, and incorporates the audio encoder from...
GLM-5.1-HighSpeed is Zhipu AI's flagship high-speed, large-model API. While retaining the full capabilities of GLM-5.1, it achieves an output speed of 400 tokens/s through the TileRT high-performance inference engine, setting a new global record...
CloudDM is an open-source database development and management platform from the ClouGence team. CloudDM integrates core database capabilities such as data querying, SQL auditing, and access control into a unified platform, supporting 30 data sources, covering MySQL...
Agora-1 is Odyssey's first multi-agent world model, enabling humans and AI to interact within the same real-time generated world simulation. The model uses the classic FPS game GoldenEye 007 as its research scenario and supports up to...
HyperEyes is a parallel multimodal search agent jointly launched by Xiaohongshu and Cambridge University. It pioneered the UGS paradigm, which integrates visual localization and retrieval into a single atomic action, enabling concurrent multi-entity search in a single round.
ZCube is a next-generation large-model inference network architecture jointly developed by Zhipu AI, Yuhun Networks, and Tsinghua University. It is specifically designed to solve structural network congestion in separate PD (Power Delivery) deployments. The architecture eliminates traditional Spine layer switches, adopting a fully flat network structure...
HiDream-O1-Image-Pro is a large-scale image model developed by Zhixiang Future, based on the native full-modal UiT architecture. With over 200 bytes of parameters, it sets state-of-the-art (SOTA) performance in tasks such as text-to-image rendering, text rendering, and instruction editing. The model integrates image pixels, text...
Lance is a lightweight, native unified multimodal model open-sourced by ByteDance's intelligent creation team. It uses only 3B activation parameters and supports the entire chain of tasks—image and video understanding, generation, and editing—within a single framework. The model employs a multimodal approach...
Gemini 3.5 Flash is Google's next-generation AI model, focusing on cutting-edge intelligence and action capabilities. The model achieved a 76.2% score on the Terminal-Bench 2.1 coding benchmark and a GDPval-AA agent task score of 16...