Best Large Language Models for Hugging Face - Page 2

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.

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.

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.

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.

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.

Qwen2.5-1M, an open-source language model from the Qwen team, has been meticulously crafted to manage context lengths reaching as high as one million tokens. This version introduces two distinct model variants, namely Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M, representing a significant advancement as it is the first instance of Qwen models being enhanced to accommodate such large context lengths. In addition to this, the team has released an inference framework that is based on vLLM and incorporates sparse attention mechanisms, which greatly enhance the processing speed for 1M-token inputs, achieving improvements between three to seven times. A detailed technical report accompanies this release, providing in-depth insights into the design choices and the results from various ablation studies. This transparency allows users to fully understand the capabilities and underlying technology of the models.

Yi-Large is an innovative proprietary large language model created by 01.AI, featuring an impressive context length of 32k and a cost structure of $2 for each million tokens for both inputs and outputs. Renowned for its superior natural language processing abilities, common-sense reasoning, and support for multiple languages, it competes effectively with top models such as GPT-4 and Claude3 across various evaluations. This model is particularly adept at handling tasks that involve intricate inference, accurate prediction, and comprehensive language comprehension, making it ideal for applications such as knowledge retrieval, data categorization, and the development of conversational chatbots that mimic human interaction. Built on a decoder-only transformer architecture, Yi-Large incorporates advanced features like pre-normalization and Group Query Attention, and it has been trained on an extensive, high-quality multilingual dataset to enhance its performance. The model's flexibility and economical pricing position it as a formidable player in the artificial intelligence landscape, especially for businesses looking to implement AI technologies on a global scale. Additionally, its ability to adapt to a wide range of use cases underscores its potential to revolutionize how organizations leverage language models for various needs.

Canopy Labs has unveiled Orpheus, an innovative suite of advanced speech large language models (LLMs) aimed at achieving human-like speech generation capabilities. Utilizing the Llama-3 architecture, these models have been trained on an extensive dataset comprising over 100,000 hours of English speech, allowing them to generate speech that exhibits natural intonation, emotional depth, and rhythmic flow that outperforms existing high-end closed-source alternatives. Orpheus also features zero-shot voice cloning, enabling users to mimic voices without any need for prior fine-tuning, and provides easy-to-use tags for controlling emotion and intonation. The models are engineered for low latency, achieving approximately 200ms streaming latency for real-time usage, which can be further decreased to around 100ms when utilizing input streaming. Canopy Labs has made available both pre-trained and fine-tuned models with 3 billion parameters under the flexible Apache 2.0 license, with future intentions to offer smaller models with 1 billion, 400 million, and 150 million parameters to cater to devices with limited resources. This strategic move is expected to broaden accessibility and application potential across various platforms and use cases.

The MiniMax‑M1 model, introduced by MiniMax AI and licensed under Apache 2.0, represents a significant advancement in hybrid-attention reasoning architecture. With an extraordinary capacity for handling a 1 million-token context window and generating outputs of up to 80,000 tokens, it facilitates in-depth analysis of lengthy texts. Utilizing a cutting-edge CISPO algorithm, MiniMax‑M1 was trained through extensive reinforcement learning, achieving completion on 512 H800 GPUs in approximately three weeks. This model sets a new benchmark in performance across various domains, including mathematics, programming, software development, tool utilization, and understanding of long contexts, either matching or surpassing the capabilities of leading models in the field. Additionally, users can choose between two distinct variants of the model, each with a thinking budget of either 40K or 80K, and access the model's weights and deployment instructions on platforms like GitHub and Hugging Face. Such features make MiniMax‑M1 a versatile tool for developers and researchers alike.

LFM2 represents an advanced series of on-device foundation models designed to provide a remarkably swift generative-AI experience across a diverse array of devices. By utilizing a novel hybrid architecture, it achieves decoding and pre-filling speeds that are up to twice as fast as those of similar models, while also enhancing training efficiency by as much as three times compared to its predecessor. These models offer a perfect equilibrium of quality, latency, and memory utilization suitable for embedded system deployment, facilitating real-time, on-device AI functionality in smartphones, laptops, vehicles, wearables, and various other platforms, which results in millisecond inference, device durability, and complete data sovereignty. LFM2 is offered in three configurations featuring 0.35 billion, 0.7 billion, and 1.2 billion parameters, showcasing benchmark results that surpass similarly scaled models in areas including knowledge recall, mathematics, multilingual instruction adherence, and conversational dialogue assessments. With these capabilities, LFM2 not only enhances user experience but also sets a new standard for on-device AI performance.

Qwen3.6 is an advanced AI model from Alibaba that builds on previous Qwen releases with a focus on real-world utility and performance. It is designed as a multimodal large language model capable of understanding and generating text while also processing visual and structured data. The model is optimized for coding tasks, enabling developers to handle complex, repository-level programming workflows. Qwen3.6 uses a mixture-of-experts (MoE) architecture, which activates only a portion of its parameters during inference to improve efficiency. This design allows it to deliver strong performance while reducing computational costs. It is available in both proprietary and open-weight versions, giving developers flexibility in deployment. The model supports integration into enterprise systems and cloud platforms, particularly within Alibaba’s ecosystem. Qwen3.6 also introduces stronger agentic capabilities, allowing it to perform multi-step reasoning and more autonomous task execution. It is designed to handle complex workflows, including engineering, analysis, and decision-making tasks. The model emphasizes stability and responsiveness based on developer feedback. Overall, Qwen3.6 provides a scalable and efficient AI solution for coding, automation, and multimodal applications.

The Gemma family consists of advanced, lightweight models developed using the same innovative research and technology as the Gemini models. These cutting-edge models are equipped with robust security features that promote responsible and trustworthy AI applications, achieved through carefully curated data sets and thorough refinements. Notably, Gemma models excel in their various sizes—2B, 7B, 9B, and 27B—often exceeding the performance of some larger open models. With the introduction of Keras 3.0, users can experience effortless integration with JAX, TensorFlow, and PyTorch, providing flexibility in framework selection based on specific tasks. Designed for peak performance and remarkable efficiency, Gemma 2 is specifically optimized for rapid inference across a range of hardware platforms. Furthermore, the Gemma family includes diverse models that cater to distinct use cases, ensuring they adapt effectively to user requirements. These lightweight language models feature a decoder and have been trained on an extensive array of textual data, programming code, and mathematical concepts, which enhances their versatility and utility in various applications.

Jamba stands out as the most potent and effective long context model, specifically designed for builders while catering to enterprise needs. With superior latency compared to other leading models of similar sizes, Jamba boasts a remarkable 256k context window, the longest that is openly accessible. Its innovative Mamba-Transformer MoE architecture focuses on maximizing cost-effectiveness and efficiency. Key features available out of the box include function calls, JSON mode output, document objects, and citation mode, all designed to enhance user experience. Jamba 1.5 models deliver exceptional performance throughout their extensive context window and consistently achieve high scores on various quality benchmarks. Enterprises can benefit from secure deployment options tailored to their unique requirements, allowing for seamless integration into existing systems. Jamba can be easily accessed on our robust SaaS platform, while deployment options extend to strategic partners, ensuring flexibility for users. For organizations with specialized needs, we provide dedicated management and continuous pre-training, ensuring that every client can leverage Jamba’s capabilities to the fullest. This adaptability makes Jamba a prime choice for enterprises looking for cutting-edge solutions.

Phi-4 is an advanced small language model (SLM) comprising 14 billion parameters, showcasing exceptional capabilities in intricate reasoning tasks, particularly in mathematics, alongside typical language processing functions. As the newest addition to the Phi family of small language models, Phi-4 illustrates the potential advancements we can achieve while exploring the limits of SLM technology. It is currently accessible on Azure AI Foundry under a Microsoft Research License Agreement (MSRLA) and is set to be released on Hugging Face in the near future. Due to significant improvements in processes such as the employment of high-quality synthetic datasets and the careful curation of organic data, Phi-4 surpasses both comparable and larger models in mathematical reasoning tasks. This model not only emphasizes the ongoing evolution of language models but also highlights the delicate balance between model size and output quality. As we continue to innovate, Phi-4 stands as a testament to our commitment to pushing the boundaries of what's achievable within the realm of small language models.

Z.ai has unveiled its latest flagship model, GLM-4.5, which boasts an impressive 355 billion total parameters (with 32 billion active) and is complemented by the GLM-4.5-Air variant, featuring 106 billion total parameters (12 billion active), designed to integrate sophisticated reasoning, coding, and agent-like functions into a single framework. This model can switch between a "thinking" mode for intricate, multi-step reasoning and tool usage and a "non-thinking" mode that facilitates rapid responses, accommodating a context length of up to 128K tokens and enabling native function invocation. Accessible through the Z.ai chat platform and API, and with open weights available on platforms like HuggingFace and ModelScope, GLM-4.5 is adept at processing a wide range of inputs for tasks such as general problem solving, common-sense reasoning, coding from the ground up or within existing frameworks, as well as managing comprehensive workflows like web browsing and slide generation. The architecture is underpinned by a Mixture-of-Experts design, featuring loss-free balance routing, grouped-query attention mechanisms, and an MTP layer that facilitates speculative decoding, ensuring it meets enterprise-level performance standards while remaining adaptable to various applications. As a result, GLM-4.5 sets a new benchmark for AI capabilities across numerous domains.

Qwen3.7-Plus is an advanced multimodal agent model that seamlessly integrates vision and language into a single, adaptable foundation for intelligent agents. Expanding upon the agentic intelligence of Qwen3.7, it enhances its abilities to include visual comprehension, reasoning, grounded interactions, and the use of various multimodal tools, allowing agents to perceive, analyze, and operate within text, images, documents, screens, and intricate real-world scenarios. This model is specifically crafted for dynamic tasks that go beyond mere static question answering, facilitating activities such as visual searches, document understanding, chart and table evaluations, screen comprehension, GUI interactions, image-driven reasoning, and workflows where perception, planning, and action are interlinked. Qwen3.7-Plus fortifies the relationship between linguistic reasoning and visual cues, empowering users to inquire about images, decode complex multimodal information, extract organized data, and formulate responses that incorporate both contextual and visual elements, thus broadening the scope of interactive AI applications. With these enhancements, users can engage in more sophisticated and nuanced interactions with the system, making it a powerful tool for various practical applications.

Qwen3.8-2.4T-A95B stands out as the most extensive open model within the Qwen3.8 series, offering advanced Qwen-Max-class features in a publicly accessible format. Constructed upon the solid framework of Qwen3.5, this model significantly enhances performance in areas such as coding, professional tasks, research, and complex, prolonged agentic activities, emphasizing the reliability of executing intricate, multi-step workflows to completion. Utilizing a cutting-edge mixture-of-experts architecture, it boasts an impressive total of 2.4 trillion parameters, with 95 billion of those being activated, featuring 512 experts and engaging 10 routed along with one shared expert simultaneously. The model accommodates a native context length of 262,144 tokens, which can be extended to around 1.01 million tokens, thereby providing substantial flexibility for various applications. Furthermore, improvements in agent execution, such as enhanced autonomous planning and better responsiveness to environmental feedback, contribute to its efficiency, while its broader compatibility with widely used agent frameworks and development tools facilitates seamless integration into existing systems, making it a versatile choice for developers and researchers alike.

Harness the capabilities of tailored models to gain a strategic edge in your market. With our advanced enterprise Gen AI framework, you can surpass conventional limits and delegate repetitive tasks to robust assistants in real time – the possibilities are endless. For businesses that prioritize data protection, customize and implement generative AI solutions within your own secure cloud environment, ensuring safety and confidentiality at every step.