Business Software for Hugging Face

The Google AI Edge Gallery is an innovative, open-source Android application designed to showcase various applications of on-device machine learning and generative AI, allowing users to download and utilize models offline once installed. This app features a range of functionalities, such as AI Chat for engaging in multi-turn conversations, Ask Image for uploading images to inquire about objects or obtain descriptions, Audio Scribe for transcribing or translating audio files, and Prompt Lab for performing single-turn tasks like summarization and code generation. Additionally, it provides performance insights, offering metrics on aspects like latency and decode speed. Users have the flexibility to switch between compatible models, including options like Gemma 3n and models from Hugging Face, as well as the ability to incorporate their own LiteRT models while accessing model cards and source code for increased transparency. By processing all data locally on the device, the app prioritizes user privacy, requiring no internet connection for core functionalities after the initial model load, which ultimately minimizes latency and bolsters data security. Overall, the Google AI Edge Gallery empowers users to explore cutting-edge AI capabilities while maintaining their privacy and control over their data.

DeepSeek has launched DeepSeek-V3.1-Terminus, an upgrade to the V3.1 architecture that integrates user suggestions to enhance output stability, consistency, and overall agent performance. This new version significantly decreases the occurrences of mixed Chinese and English characters as well as unintended distortions, leading to a cleaner and more uniform language generation experience. Additionally, the update revamps both the code agent and search agent subsystems to deliver improved and more dependable performance across various benchmarks. DeepSeek-V3.1-Terminus is available as an open-source model, with its weights accessible on Hugging Face, making it easier for the community to leverage its capabilities. The structure of the model remains consistent with DeepSeek-V3, ensuring it is compatible with existing deployment strategies, and updated inference demonstrations are provided for users to explore. Notably, the model operates at a substantial scale of 685B parameters and supports multiple tensor formats, including FP8, BF16, and F32, providing adaptability in different environments. This flexibility allows developers to choose the most suitable format based on their specific needs and resource constraints.

Introducing DeepSeek-V3.2-Exp, our newest experimental model derived from V3.1-Terminus, featuring the innovative DeepSeek Sparse Attention (DSA) that enhances both training and inference speed for lengthy contexts. This DSA mechanism allows for precise sparse attention while maintaining output quality, leading to improved performance for tasks involving long contexts and a decrease in computational expenses. Benchmark tests reveal that V3.2-Exp matches the performance of V3.1-Terminus while achieving these efficiency improvements. The model is now fully operational across app, web, and API platforms. Additionally, to enhance accessibility, we have slashed DeepSeek API prices by over 50% effective immediately. During a transition period, users can still utilize V3.1-Terminus via a temporary API endpoint until October 15, 2025. DeepSeek encourages users to share their insights regarding DSA through our feedback portal. Complementing the launch, DeepSeek-V3.2-Exp has been made open-source, with model weights and essential technology—including crucial GPU kernels in TileLang and CUDA—accessible on Hugging Face. We look forward to seeing how the community engages with this advancement.

Tiger Data reimagines PostgreSQL for the modern era — powering everything from IoT and fintech to AI and Web3. As the creator of TimescaleDB, it brings native time-series, event, and analytical capabilities to the world’s most trusted database engine. Through Tiger Cloud, developers gain access to a fully managed, elastic infrastructure with auto-scaling, high availability, and point-in-time recovery. The platform introduces core innovations like Forks (copy-on-write storage branches for CI/CD and testing), Memory (durable agent context and recall), and Search (hybrid BM25 and vector retrieval). Combined with hypertables, continuous aggregates, and materialized views, Tiger delivers the speed of specialized analytical systems without sacrificing SQL simplicity. Teams use Tiger Data to unify real-time and historical analytics, build AI-driven workflows, and streamline data management at scale. It integrates seamlessly with the entire PostgreSQL ecosystem, supporting APIs, CLIs, and modern development frameworks. With over 20,000 GitHub stars and a thriving developer community, Tiger Data stands as the evolution of PostgreSQL for the intelligent data age.

BotDojo serves as a robust AI enablement platform tailored for enterprises, allowing companies to create, implement, oversee, and expand intelligent agents across various communication channels like chat, voice, email, and web, all through an intuitive low-code visual workflow designer that seamlessly integrates with existing enterprise data systems. It boasts a library of over 100 pre-built templates aimed at streamlining typical applications, including support automation, knowledge retrieval, sales analytics, and internal operations, while also facilitating branching logic, memory capabilities, and the orchestration of tools such as code, RPA, and web browsing. In addition, BotDojo establishes connections with essential business tools like CRMs, ticketing platforms, and databases to enhance its functionality. The platform further fosters continuous improvement and learning for agents through human feedback loops, enabling employees to mentor agents by providing feedback, embedding corrections into agent memory and responses, and assessing performance using comprehensive observability metrics, including deflection rates, first-contact resolution, and cost per interaction. Ultimately, BotDojo not only optimizes operational efficiency but also ensures that intelligent agents evolve and adapt to meet organizational needs effectively.

Ekinox serves as a visual AI automation platform that allows users to create, implement, and oversee AI-driven workflows without the need for coding; its user-friendly drag-and-drop interface facilitates the design of intelligent agents that can link to over 100 pre-existing integrations, triggering actions across numerous productivity, data, and communication applications. The platform is designed for real-time processing and encourages collaboration by offering team workspaces, version control, and immediate deployment capabilities. In addition, it boasts enterprise-level security that adheres to SOC 2 standards, features bank-level encryption, supports custom API connectors, and includes sophisticated access controls. Users benefit from the ability to monitor their workflows through comprehensive analytics dashboards, enabling them to assess costs and performance across various models and integrations while utilizing predictive auto-scaling and log retention for enhanced functionality. With setup times cut down to mere minutes, Ekinox optimizes processes ranging from straightforward task automation to more complex workflows, making it an invaluable tool. This efficiency not only improves productivity but also enhances the overall user experience.

RenderFlow AI is a cloud-based platform that generates animated videos of professional quality from simple text prompts or uploaded images, utilizing various AI models. Users are able to articulate scenes using natural language, choose their preferred style and model, and modify factors such as duration and resolution, after which the system generates a refined final product, complete with commercial usage rights. Prioritizing rapid production, it claims to deliver videos in mere minutes, contrasting sharply with the protracted processes typical of traditional editing methods, and is versatile enough to cater to different needs such as product demonstrations, animated visual content, social media posts, and educational videos. The user-friendly interface and flexibility in model selection, combined with assertions of producing high-quality results even for those without expertise, ensure that it serves as an accessible video creation solution for both industry professionals and everyday users alike. This makes it an appealing option for anyone looking to create compelling visual narratives with minimal effort.

DeepSeek-V3.2 is a highly optimized large language model engineered to balance top-tier reasoning performance with significant computational efficiency. It builds on DeepSeek's innovations by introducing DeepSeek Sparse Attention (DSA), a custom attention algorithm that reduces complexity and excels in long-context environments. The model is trained using a sophisticated reinforcement learning approach that scales post-training compute, enabling it to perform on par with GPT-5 and match the reasoning skill of Gemini-3.0-Pro. Its Speciale variant overachieves in demanding reasoning benchmarks and does not include tool-calling capabilities, making it ideal for deep problem-solving tasks. DeepSeek-V3.2 is also trained using an agentic synthesis pipeline that creates high-quality, multi-step interactive data to improve decision-making, compliance, and tool-integration skills. It introduces a new chat template design featuring explicit thinking sections, improved tool-calling syntax, and a dedicated developer role used strictly for search-agent workflows. Users can encode messages using provided Python utilities that convert OpenAI-style chat messages into the expected DeepSeek format. Fully open-source under the MIT license, DeepSeek-V3.2 is a flexible, cutting-edge model for researchers, developers, and enterprise AI teams.

DeepSeek-V3.2-Speciale is the most advanced reasoning-focused version of the DeepSeek-V3.2 family, designed to excel in mathematical, algorithmic, and logic-intensive tasks. It incorporates DeepSeek Sparse Attention (DSA), an efficient attention mechanism tailored for very long contexts, enabling scalable reasoning with minimal compute costs. The model undergoes a robust reinforcement learning pipeline that scales post-training compute to frontier levels, enabling performance that exceeds GPT-5 on internal evaluations. Its achievements include gold-medal-level solutions in IMO 2025, IOI 2025, ICPC World Finals, and CMO 2025, with final submissions publicly released for verification. Unlike the standard V3.2 model, the Speciale variant removes tool-calling capabilities to maximize focused reasoning output without external interactions. DeepSeek-V3.2-Speciale uses a revised chat template with explicit thinking blocks and system-level reasoning formatting. The repository includes encoding tools showing how to convert OpenAI-style chat messages into DeepSeek’s specialized input format. With its MIT license and 685B-parameter architecture, DeepSeek-V3.2-Speciale offers cutting-edge performance for academic research, competitive programming, and enterprise-level reasoning applications.

DeepCoder, an entirely open-source model for code reasoning and generation, has been developed through a partnership between Agentica Project and Together AI. Leveraging the foundation of DeepSeek-R1-Distilled-Qwen-14B, it has undergone fine-tuning via distributed reinforcement learning, achieving a notable accuracy of 60.6% on LiveCodeBench, which marks an 8% enhancement over its predecessor. This level of performance rivals that of proprietary models like o3-mini (2025-01-031 Low) and o1, all while operating with only 14 billion parameters. The training process spanned 2.5 weeks on 32 H100 GPUs, utilizing a carefully curated dataset of approximately 24,000 coding challenges sourced from validated platforms, including TACO-Verified, PrimeIntellect SYNTHETIC-1, and submissions to LiveCodeBench. Each problem mandated a legitimate solution along with a minimum of five unit tests to guarantee reliability during reinforcement learning training. Furthermore, to effectively manage long-range context, DeepCoder incorporates strategies such as iterative context lengthening and overlong filtering, ensuring it remains adept at handling complex coding tasks. This innovative approach allows DeepCoder to maintain high standards of accuracy and reliability in its code generation capabilities.

MiMo-V2-Flash is a large language model created by Xiaomi that utilizes a Mixture-of-Experts (MoE) framework, combining remarkable performance with efficient inference capabilities. With a total of 309 billion parameters, it activates just 15 billion parameters during each inference, allowing it to effectively balance reasoning quality and computational efficiency. This model is well-suited for handling lengthy contexts, making it ideal for tasks such as long-document comprehension, code generation, and multi-step workflows. Its hybrid attention mechanism integrates both sliding-window and global attention layers, which helps to minimize memory consumption while preserving the ability to understand long-range dependencies. Additionally, the Multi-Token Prediction (MTP) design enhances inference speed by enabling the simultaneous processing of batches of tokens. MiMo-V2-Flash boasts impressive generation rates of up to approximately 150 tokens per second and is specifically optimized for applications that demand continuous reasoning and multi-turn interactions. The innovative architecture of this model reflects a significant advancement in the field of language processing.

TranslateGemma is an innovative collection of open machine translation models created by Google, based on the Gemma 3 architecture, which facilitates communication between individuals and systems in 55 languages by providing high-quality AI translations while ensuring efficiency and wide deployment options. Offered in sizes of 4 B, 12 B, and 27 B parameters, TranslateGemma encapsulates sophisticated multilingual functionalities into streamlined models that are capable of functioning on mobile devices, consumer laptops, local systems, or cloud infrastructure, all without compromising on precision or performance; assessments indicate that the 12 B variant can exceed the capabilities of larger baseline models while requiring less computational power. The development of these models involved a distinct two-phase fine-tuning approach that integrates high-quality human and synthetic translation data, using reinforcement learning to enhance translation accuracy across a variety of language families. This innovative methodology ensures that users benefit from an array of languages while experiencing swift and reliable translations.

Liquid AI's LFM2.5 represents an advanced iteration of on-device AI foundation models, engineered to provide high-efficiency and performance for AI inference on edge devices like smartphones, laptops, vehicles, IoT systems, and embedded hardware without the need for cloud computing resources. This new version builds upon the earlier LFM2 framework by greatly enhancing the scale of pretraining and the stages of reinforcement learning, resulting in a suite of hybrid models that boast around 1.2 billion parameters while effectively balancing instruction adherence, reasoning skills, and multimodal functionalities for practical applications. The LFM2.5 series comprises various models including Base (for fine-tuning and personalization), Instruct (designed for general-purpose instruction), Japanese-optimized, Vision-Language, and Audio-Language variants, all meticulously crafted for rapid on-device inference even with stringent memory limitations. These models are also made available as open-weight options, facilitating deployment through platforms such as llama.cpp, MLX, vLLM, and ONNX, thus ensuring versatility for developers. With these enhancements, LFM2.5 positions itself as a robust solution for diverse AI-driven tasks in real-world environments.

Step 3.5 Flash is a cutting-edge open-source foundational language model designed for advanced reasoning and agent-like capabilities, optimized for efficiency; it utilizes a sparse Mixture of Experts (MoE) architecture that activates only approximately 11 billion of its nearly 196 billion parameters per token, ensuring high-density intelligence and quick responsiveness. The model features a 3-way Multi-Token Prediction (MTP-3) mechanism that allows it to generate hundreds of tokens per second, facilitating complex multi-step reasoning and task execution while efficiently managing long contexts through a hybrid sliding window attention method that minimizes computational demands across extensive datasets or codebases. Its performance on reasoning, coding, and agentic tasks is formidable, often matching or surpassing that of much larger proprietary models, and it incorporates a scalable reinforcement learning system that enables continuous self-enhancement. Moreover, this innovative approach positions Step 3.5 Flash as a significant player in the field of AI language models, showcasing its potential to revolutionize various applications.

Agent Zero is an innovative open source framework for AI agents that enables the development of autonomous assistants capable of executing intricate tasks through direct interaction with computer systems. This platform offers a unique setting where AI agents can access real system functions, empowering them to run commands, write and execute code, navigate the internet, analyze data, and oversee workflows as part of comprehensive automation solutions. Unlike a standard chat interface, Agent Zero operates within its isolated virtual environment, enabling it to engage with the operating system, install necessary tools, run scripts, and manage tasks across various components seamlessly. The framework prioritizes transparency and developer control, allowing users to monitor, adjust, and personalize agent behavior, tool accessibility, and information processing methods. With a modular architecture, Agent Zero facilitates the dynamic creation and utilization of tools, all while maintaining a consistent memory for enhanced performance. This makes it an ideal choice for developers aiming to build highly customizable and efficient AI-driven workflows.

Subconscious is a platform tailored for developers that simplifies the creation, deployment, and scaling of production-ready AI agents by automating the most challenging aspects of agent architecture. By offering a comprehensive agent system, it takes care of context management, tool orchestration, and facilitates long-term reasoning, allowing developers to concentrate on setting objectives and defining functionalities instead of dealing with intricate infrastructure setups. The platform features a cohesive inference engine that combines a jointly designed model and runtime, enabling the breakdown of complex tasks, dynamic workflow generation, and the execution of multi-step reasoning without the need for manual context management or coordination among multiple agents. In contrast to conventional methods that depend on linking various APIs and frameworks, Subconscious empowers agents to receive goals and tools and then independently plan, reason, and act with minimal human oversight. This innovation effectively results in systems that can autonomously accomplish tasks, streamlining the development process for AI applications. As a result, developers can realize their visions more efficiently and with greater ease.

Atomic Chat is an innovative conversational platform powered by artificial intelligence, designed to streamline and automate customer interactions across various messaging channels, which allows businesses to connect, qualify, and convert leads through immediate engagement. By consolidating conversations from popular platforms like WhatsApp, Messenger, Instagram, and Telegram into one comprehensive inbox, teams can efficiently oversee all customer communications while ensuring complete visibility and control. The platform employs intelligent AI agents capable of managing conversations through text, voice, and image inputs, delivering human-like responses that can address inquiries, qualify leads, schedule meetings, and conduct follow-ups automatically, around the clock. Additionally, it facilitates the automation of customer service workflows and sales strategies, such as lead scoring, re-engagement campaigns, and tailored messaging sequences, which enhance conversion rates and alleviate manual efforts. Consequently, businesses can focus more on strategic initiatives while the platform handles routine interactions seamlessly.

Better Auth is a versatile authentication and authorization solution tailored for TypeScript, enabling developers to seamlessly integrate secure login functionalities into their applications and databases. It comes equipped with a comprehensive suite of authentication tools right from the start, offering features such as email and password login, session management, email verification, password resets, and compatibility with over 40 social login options like Google and GitHub, all requiring minimal coding effort. Designed to be compatible with a broad spectrum of contemporary frameworks, including Next.js, Nuxt, SvelteKit, Astro, and Express, it empowers teams to implement authentication irrespective of their chosen technology stack while ensuring robust TypeScript support and type safety. Furthermore, Better Auth boasts sophisticated features such as multi-factor authentication, management of multi-tenant organizations, and enterprise-level functionalities including SSO, SAML, and SCIM provisioning, making it an ideal fit for both straightforward applications and complex, large-scale systems. This flexibility allows developers to focus on building their core application features while relying on Better Auth to handle security concerns efficiently.

Locally AI is an innovative application that empowers users to utilize advanced language models directly on their iPhone, iPad, or Mac without needing cloud services or an internet connection. Leveraging Apple’s MLX framework, it provides quick and efficient performance while keeping power consumption low, thus ensuring a fluid experience for chatting, creating, learning, and discovering AI capabilities across various devices. The app supports a range of open models, including Llama, Gemma, Qwen, and DeepSeek, enabling users to easily switch between them and customize outputs for various tasks. Operating entirely offline, it eliminates the need for logins and ensures that no data is collected or transmitted, thereby guaranteeing complete privacy and control over personal information. Users can engage with AI through natural dialogue, assess documents or images, and produce text within a user-friendly interface that prioritizes simplicity and responsiveness. This design fosters greater creativity and exploration, further enhancing the overall user experience.

Qwen3.5-35B-A3B is a member of the Qwen3.5 "Medium" model series, meticulously crafted as an effective multimodal foundation model that strikes a balance between robust reasoning capabilities and practical application needs. Utilizing a Mixture-of-Experts (MoE) architecture, it boasts a total of 35 billion parameters, yet activates only around 3 billion for each token, enabling it to achieve performance levels similar to much larger models while significantly cutting down on computational expenses. The model employs a hybrid attention mechanism that merges linear attention with traditional attention layers, which enhances its ability to handle extensive context and boosts scalability for intricate tasks. As an inherently vision-language model, it processes both textual and visual data, catering to a variety of applications, including multimodal reasoning, programming, and automated workflows. Furthermore, it is engineered to operate as a versatile "AI agent," proficient in planning, utilizing tools, and systematically solving problems, extending its functionality beyond mere conversational interactions. This capability positions it as a valuable asset across diverse domains, where advanced AI-driven solutions are increasingly required.

Qwen3.6-27B is an open-source, dense multimodal language model from the Qwen3.6 series, engineered to provide top-tier performance in areas such as coding, reasoning, and agent-driven workflows, all while maintaining an efficient parameter count of 27 billion. This model is recognized for its ability to outperform or compete closely with much larger counterparts on essential benchmarks, particularly excelling in agent-based coding tasks. It features dual operational modes—thinking and non-thinking—that enable it to effectively adapt its reasoning depth and response speed based on the specific requirements of each task. Additionally, it supports a variety of input types, including text, images, and video, showcasing its versatility. As part of the Qwen3.6 lineup, this model prioritizes practical usability, consistency, and the enhancement of developer productivity, reflecting advancements inspired by community insights and real-world application demands. Its innovative design not only responds to immediate user needs but also anticipates future trends in AI development.

Sarvam-30B is an advanced open-source large language model that serves as a comprehensive platform for real-time conversational AI and complex reasoning tasks, emphasizing its capability in multilingual settings and practical usage. This 30-billion parameter model is engineered for enhanced speed and efficiency through a Mixture-of-Experts (MoE) framework, which selectively activates a portion of its parameters for each request, thus facilitating high throughput and minimal latency while remaining suitable for environments with limited resources, including local devices and edge computing systems. It excels in various conversational applications, programming tasks, and logical reasoning, achieving impressive outcomes in over 20 Indian languages, which underscores its utility for multilingual applications and voice interaction systems. The model features a dual-tier structure, acting as a rapid and deployable "conversational workhorse," and utilizes MoE techniques to lower computational costs without sacrificing performance. This innovative model not only enhances user experience but also broadens accessibility in diverse linguistic contexts.

Sarvam-105B stands as the premier large language model within Sarvam’s open-source lineup, engineered to provide exceptional reasoning capabilities, multilingual comprehension, and agent-driven execution all within a unified and scalable framework. This Mixture-of-Experts (MoE) model boasts an impressive total of approximately 105 billion parameters, activating only a subset for each token, which allows it to maintain superior computational efficiency while excelling in intricate tasks. It is particularly optimized for advanced reasoning, programming, mathematical challenges, and agentic processes, positioning it well for scenarios that necessitate multi-step problem-solving and organized outputs rather than merely engaging in basic conversations. With the ability to process long contexts of around 128K tokens, Sarvam-105B can effectively manage extensive documents, prolonged discussions, and complex analytical inquiries, ensuring coherence throughout. Additionally, its design facilitates a diverse range of applications, providing users with versatile tools to tackle a variety of intellectual challenges.

Trismik serves as a platform for evaluating AI models, aimed at assisting teams in selecting the most suitable large language model tailored to their unique needs by utilizing actual data rather than mere assumptions or standard benchmarks. The platform emphasizes transforming the process of model experimentation into straightforward, evidence-based choices by giving users the ability to test and contrast various models directly with their own datasets, avoiding the pitfalls of public leaderboards or limited manual evaluations. Alongside this, it features innovative tools like QuickCompare, which allows for side-by-side assessments of over 50 models across essential metrics such as quality, cost, and speed, thus rendering trade-offs visible and quantifiable in practical scenarios. Additionally, Trismik employs adaptive evaluation methods inspired by psychometrics, which intelligently select the most informative test cases and automatically assess outputs across multiple dimensions, including factual accuracy, bias, and reliability, ensuring a comprehensive evaluation process. This holistic approach not only enhances the decision-making process but also empowers teams to make informed choices that align with their specific operational requirements.

Pi is a streamlined terminal coding environment designed to seamlessly integrate with developer workflows rather than requiring developers to conform to its structure. It comes equipped with robust default settings while maintaining a compact size and extensive customization options, allowing users to enhance Pi through various extensions, skills, prompt templates, themes, and shareable packages sourced from npm or git. When a team requires a specific command, tool, provider, workflow, or UI modification, they can simply instruct Pi to create it, make adjustments on the fly, reload, and continue their work without interruption. Pi is versatile, offering support for interactive, print/JSON, RPC, and SDK modes, which enables it to function as a comprehensive terminal UI, a scriptable command interface, a JSON event stream, or an easily embeddable agent harness. It is compatible with over 15 providers and numerous models, including options like Anthropic, OpenAI, Google, Azure, Bedrock, Mistral, Groq, Cerebras, xAI, Hugging Face, Kimi For Coding, MiniMax, OpenRouter, Ollama, and other services, facilitating mid-session model switching to enhance flexibility and user experience. This adaptability makes Pi an invaluable tool for developers looking to tailor their coding environment to meet their specific needs.