# Open-source

Kimi K2 AI is a powerful open-source chat platform with an autonomous AI agent. It outperforms GPT-4 in programming and math benchmark tests, providing enterprise-grade AI solutions with a 95% cost reduction. Kimi K2 AI is committed to delivering an efficient and intelligent chat experience, which can be widely applied in various scenarios.

Minion Agent is a simple yet powerful proxy framework that can interact with browsers, support deep research, automatic planning, and other functions. It is suitable for users who need to perform complex tasks and research. It provides a flexible toolkit that allows developers to easily integrate different models and tools. This framework not only improves work efficiency but also provides users with a convenient user experience, making it suitable for various scientific research and business applications. The product is open-source, allowing users to use and modify it freely.

DeepSeek-Prover-V2-671B is an advanced artificial intelligence model designed to provide strong reasoning capabilities. It is based on the latest technology and applicable to various scenarios. The model is open source, aiming to promote the democratization and popularization of AI technology, reduce technical barriers, and enable more developers and researchers to use AI technology for innovation. By using this model, users can enhance their work efficiency and advance the progress of various projects.

RepoCloud is an open-source application cloud market that allows users to easily deploy and scale their favorite open-source applications. It provides one-click deployment functionality at half the price of mainstream cloud hosts, aiming to reduce user deployment and operation costs.

DeepCoder-14B-Preview is a large language model for code reasoning based on reinforcement learning. It can handle long contexts, boasts a 60.6% pass rate, and is suitable for programming tasks and automated code generation. The model's advantage lies in its innovative training method, providing superior performance compared to other models, and being completely open-source, supporting a wide range of community applications and research.

StarVector is an advanced generative model designed to transform images and text instructions into high-quality scalable vector graphics (SVG) code. Its key advantages lie in its ability to handle complex SVG elements and excel across various graphic styles and complexities. As an open-source resource, StarVector drives innovation and efficiency in graphic design, suitable for various applications including design, illustration, and technical documentation.

OpenManus is an open-source intelligent agent project aimed at achieving functionalities similar to Manus through open-source means, but without requiring an invitation code. Developed collaboratively by multiple developers, it leverages a powerful language model and a flexible plugin system to quickly implement various complex tasks. The main advantages of OpenManus are that it is open-source, free, and easily extensible, making it suitable for developers and researchers for secondary development and research. The project background stems from a need to improve existing intelligent agent tools, with the goal of creating a completely open and easy-to-use intelligent agent platform.

Instella is a series of high-performance open-source language models developed by the AMD GenAI team, trained on AMD Instinct? MI300X GPUs. This model significantly outperforms other open-source language models of the same size and is comparable in functionality to models like Llama-3.2-3B and Qwen2.5-3B. Instella provides model weights, training code, and training data, aiming to promote the development of open-source language models. Its main advantages include high performance, open-source availability, and optimized support for AMD hardware.

Open-source DeepResearch is an open-source project aimed at replicating functionalities akin to OpenAI Deep Research using an open-source framework and tools. The project is based on the Hugging Face platform and leverages open-source large language models (LLMs) and agent frameworks to achieve complex multi-step reasoning and information retrieval through code agents and tool calls. Its main advantages are being open-source, highly customizable, and capable of continuous improvement through community contributions. The project aims to enable anyone to run intelligent agents similar to DeepResearch locally, using their preferred models, with full localization and customization.

The Exa & Deepseek Chat App is an open-source chat application designed to perform real-time web searches through Exa's API, combined with the Deepseek R1 language model for inference, to offer a more accurate chatting experience. Built with Next.js, TailwindCSS, and TypeScript, and hosted on Vercel, it allows users to access the latest web information during chats and engage in intelligent conversations powered by a robust language model. This application is freely available as open-source, suitable for both developers and business users, and can serve as a foundation for developing chat tools.

Baichuan-M1-14B is an open-source large language model developed by Baichuan Intelligent, optimized for medical scenarios. It is trained on high-quality medical and general data comprising 200 trillion tokens, covering over 20 medical departments, showcasing strong contextual understanding and long-sequence task performance. The model excels in healthcare applications and achieves state-of-the-art performance in general tasks for models of similar sizes. Its innovative model architecture and training approach enable outstanding performance in complex tasks such as medical reasoning and diagnosis, providing robust support for AI applications in healthcare.

Builder.io is a powerful visual development platform that rapidly transforms designs into code using AI technology, offering visual editing and enterprise-level CMS features. Its key advantages include efficient design-to-code conversion, flexible visual editing capabilities, and robust content management functionalities. It is suitable for teams needing to quickly develop and iterate digital products, providing strong support for marketing websites, landing pages, or multi-brand management. The platform has flexible pricing with a free trial option, making it suitable for businesses of all sizes.

Reactive Resume is a free and open-source resume generator developed by Amruth Pillai, designed to help users easily create, update, and share resumes. This project has been under continuous development for over 3 years, offering a rich set of features and high flexibility. It is licensed under the MIT open-source license, allowing users to freely use and modify the code, even hosting it via Docker. It supports multiple languages, a variety of resume templates, and customization options to meet diverse user needs. As a completely free and ad-free product, it emphasizes practicality and convenience, aiming to help job seekers stand out in the job search process.

Moondream AI is an open-source visual language model with powerful multimodal processing capabilities. It supports various quantization formats such as fp16, int8, and int4, enabling GPU and CPU optimized inference across target devices like servers, PCs, and mobile devices. Key advantages include fast, efficient, and easy deployment, under the Apache 2.0 license that permits users to freely use and modify it. Moondream AI is positioned to provide developers with a flexible and efficient AI solution suitable for a wide range of applications requiring visual and language processing abilities.

FlagCX is a scalable and adaptive cross-chip communication library supported by the Beijing Academy of Artificial Intelligence (BAAI). It is part of the FlagAI-Open open-source initiative aimed at fostering an open-source ecosystem for AI technologies. FlagCX utilizes native collective communication libraries and comprehensively supports single-chip communication across different platforms. Supported communication backends include NCCL, IXCCL, and CNCL.

YuLan-Mini is a lightweight language model developed by the AI Box team at Renmin University of China. With 240 million parameters, it achieves performance comparable to industry-leading models trained on larger datasets, despite using only 1.08 terabytes of pre-trained data. The model excels in mathematics and coding domains, and to facilitate reproducibility, the team will open-source relevant pre-training resources.

Boltz-1 is the first truly open-source biomolecular structure prediction model developed by researchers at the MIT Jameel Clinic for Health Machine Learning. Its accuracy is comparable to that of AlphaFold3. Named after the Boltzmann distribution, it serves as a probabilistic measure for describing the distribution of molecular structures. The development of Boltz-1 aims to foster innovation beyond academia and support commercial applications. Led by PhD students Jeremy Wohlwend, Gabriele Corso, and MIT Jameel Clinic researcher Saro Passaro, it was guided by MIT Electrical Engineering and Computer Science (EECS) professors Regina Barzilay and Tommi Jaakkola. Boltz-1 faced challenges regarding scale and data processing but ultimately constructed the necessary computational capacity, laying the foundation for standardizing structural biology research practices, with the potential to accelerate the creation of life-changing drugs.

MegaParse is a powerful file parser designed for large language models (LLMs) to ensure that no information is lost during the parsing process. It supports various file formats, including PDF, PowerPoint, Word documents, etc., and is open-source. The main advantages of this tool are its speed and efficiency, along with broad compatibility with different file types. MegaParse was developed by QuivrHQ and has an active community and contributors. The product is free and its source code is accessible through GitHub.

Llama-3.1-Tulu-3-8B-SFT is part of the Tülu3 model family, a leading class of instruction-following models that offers fully open-source data, code, and guidelines aimed at modern post-training techniques. The model excels not only in chat tasks but also demonstrates outstanding performance across various tasks such as MATH, GSM8K, and IFEval.

Tülu 3 is a series of open-source advanced language models that have been fine-tuned to adapt to various tasks and user needs. These models achieve complex training processes by combining elements of proprietary methods, innovative technology, and established academic research. The success of Tülu 3 is rooted in meticulous data management, rigorous experimentation, innovative methodologies, and enhanced training infrastructure. By openly sharing data, recipes, and findings, Tülu 3 aims to empower the community to explore new and innovative fine-tuning techniques.

DeepSeek-R1-Lite-Preview is an AI model focused on enhancing reasoning abilities, demonstrating impressive performance in the AIME and MATH benchmark tests. The model features a real-time transparent thought process and plans to release an open-source version and API. The reasoning capabilities of DeepSeek-R1-Lite-Preview improve steadily with longer thought durations, showcasing better performance. According to product background information, this latest offering from DeepSeek aims to enhance user work efficiency and problem-solving skills through AI technology. Currently, the product is available for free trial, with specific pricing and positioning details yet to be announced.

Qwen2.5-Coder-3B is a large language model within the Qwen2.5-Coder series, emphasizing code generation, reasoning, and debugging. Built on the robust Qwen2.5 architecture, this model significantly improves code generation, reasoning, and debugging capabilities by increasing training tokens to 5.5 trillion, incorporating source code, textual code fundamentals, synthetic data, and more. The Qwen2.5-Coder-32B has become the most advanced open-source code language model available, matching the encoding capabilities of GPT-4o. Furthermore, Qwen2.5-Coder-3B provides a comprehensive foundation for real-world applications like code assistants, enhancing coding capabilities while maintaining strengths in mathematics and general comprehension.

easegen-admin is an open-source digital course creation platform designed to contribute to AI development through open-source contributions. The frontend is built on Vue3 + element-plus, while the backend is powered by ruoyi-vue-pro. The intelligent courseware is based on Wenduoduo. It offers a comprehensive education technology product with features like the course creation page, my videos page, intelligent courseware page, and intelligent question generation page, supporting video display and intelligent question generation.

SiYuan is a personal knowledge management system centered around privacy protection, supporting block references and bidirectional linking. Written in TypeScript and Golang, it offers features such as block-level content organization, a WYSIWYG Markdown editor, content block snapshots, mind maps, back-linking, and references. SiYuan allows for nonlinear content organization, facilitates information connectivity, and aids in building a knowledge network. Moreover, SiYuan supports data self-hosting, ensuring user data security and privacy. It boasts an active community and a rich ecosystem of plugins, compatible with multiple operating systems including Windows, macOS, and Linux. Most features are free, with premium features available for a fee, such as data synchronization.

FaceFusion Labs is a leading platform focused on facial manipulation, utilizing advanced technologies to achieve facial feature fusion and manipulation. Key advantages of the platform include high-precision facial recognition and fusion capabilities, as well as a developer-friendly API interface. Background information indicates that it had its initial commit on October 15, 2024, led by Henry Ruhs. The product is positioned as an open-source project, encouraging community contributions and collaboration.

Mermaid Whiteboard is a text-based charting tool developed by the team behind the award-winning open-source project Mermaid JS. It enables users to create a variety of charts, including flowcharts, sequence diagrams, Gantt charts, and more, through text. Mermaid Charts have revolutionized system design and training for new team members by streamlining documentation processes and enhancing workflow and communication efficiency among teams.

BlinkShot is a real-time AI image generator based on Together AI, utilizing Flux technology to generate images in milliseconds as users input prompts. This product is 100% free and open-source, designed to empower creative professionals and developers with rapid image generation capabilities to support their design and creative endeavors.

Shusheng: Skyline LandMark is a real-world 3D model based on NeRF technology, achieving 4K HD training over an extensive area of 100 square kilometers, with capabilities for real-time rendering and free editing. This technology represents a new pinnacle in city-scale 3D modeling and rendering, boasting high efficiency for training and rendering, making it a robust tool for urban planning, architectural design, and virtual reality.

Daily Bots is an open-source cloud platform focused on providing ultra-low latency voice and video AI services. It enables developers to build and host agents on a real-time global infrastructure, leveraging a rapidly growing open-source real-time framework. The platform boasts a global real-time cloud with a 13 ms first-hop latency and supports 500 million end-users, complying with SOC 2, HIPAA, and GDPR standards. Moreover, Daily Bots offers a comprehensive enterprise connection solution for telephony and workflows, complete with PSTN and SIP stacks.

Datalab's AI for Document Intelligence includes a suite of AI models for intelligent document processing, such as OCR, layout analysis, and PDF to Markdown conversion. These models represent the latest advancements in document processing technology, are user-friendly, and open-source, making them highly applicable for enhancing the efficiency and accuracy of document handling.