Alternatives to GPT-NeoX

Large language models, often requiring extensive computational resources for training over long periods, have demonstrated impressive proficiency in zero- and few-shot learning tasks. Due to the high investment needed for their development, replicating these models poses a significant challenge for many researchers. Furthermore, access to the few models available via API is limited, as users cannot obtain the complete model weights, complicating academic exploration. In response to this, we introduce Open Pre-trained Transformers (OPT), a collection of decoder-only pre-trained transformers ranging from 125 million to 175 billion parameters, which we intend to share comprehensively and responsibly with interested scholars. Our findings indicate that OPT-175B exhibits performance on par with GPT-3, yet it is developed with only one-seventh of the carbon emissions required for GPT-3's training. Additionally, we will provide a detailed logbook that outlines the infrastructure hurdles we encountered throughout the project, as well as code to facilitate experimentation with all released models, ensuring that researchers have the tools they need to explore this technology further.

GPT-J represents an advanced language model developed by EleutherAI, known for its impressive capabilities. When it comes to performance, GPT-J showcases a proficiency that rivals OpenAI's well-known GPT-3 in various zero-shot tasks. Remarkably, it has even outperformed GPT-3 in specific areas, such as code generation. The most recent version of this model, called GPT-J-6B, is constructed using a comprehensive linguistic dataset known as The Pile, which is publicly accessible and consists of an extensive 825 gibibytes of language data divided into 22 unique subsets. Although GPT-J possesses similarities to ChatGPT, it's crucial to highlight that it is primarily intended for text prediction rather than functioning as a chatbot. In a notable advancement in March 2023, Databricks unveiled Dolly, a model that is capable of following instructions and operates under an Apache license, further enriching the landscape of language models. This evolution in AI technology continues to push the boundaries of what is possible in natural language processing.

We introduce T5, a model that transforms all natural language processing tasks into a consistent text-to-text format, ensuring that both inputs and outputs are text strings, unlike BERT-style models which are limited to providing either a class label or a segment of the input text. This innovative text-to-text approach enables us to utilize the same model architecture, loss function, and hyperparameter settings across various NLP tasks such as machine translation, document summarization, question answering, and classification, including sentiment analysis. Furthermore, T5's versatility extends to regression tasks, where it can be trained to output the textual form of a number rather than the number itself, showcasing its adaptability. This unified framework greatly simplifies the handling of diverse NLP challenges, promoting efficiency and consistency in model training and application.

Pythia integrates the examination of interpretability and scaling principles to gain insights into the progression and transformation of knowledge throughout the training of autoregressive transformer models. This approach enables a deeper understanding of the mechanisms behind model learning and adaptation.

Seed Diffusion Preview is an advanced language model designed for code generation that employs discrete-state diffusion, allowing it to produce code in a non-sequential manner, resulting in significantly faster inference times without compromising on quality. This innovative approach utilizes a two-stage training process that involves mask-based corruption followed by edit-based augmentation, enabling a standard dense Transformer to achieve an optimal balance between speed and precision while avoiding shortcuts like carry-over unmasking, which helps maintain rigorous density estimation. The model impressively achieves an inference rate of 2,146 tokens per second on H20 GPUs, surpassing current diffusion benchmarks while either matching or exceeding their accuracy on established code evaluation metrics, including various editing tasks. This performance not only sets a new benchmark for the speed-quality trade-off in code generation but also showcases the effective application of discrete diffusion methods in practical coding scenarios. Its success opens up new avenues for enhancing efficiency in coding tasks across multiple platforms.

DeepSpeed is an open-source library focused on optimizing deep learning processes for PyTorch. Its primary goal is to enhance efficiency by minimizing computational power and memory requirements while facilitating the training of large-scale distributed models with improved parallel processing capabilities on available hardware. By leveraging advanced techniques, DeepSpeed achieves low latency and high throughput during model training. This tool can handle deep learning models with parameter counts exceeding one hundred billion on contemporary GPU clusters, and it is capable of training models with up to 13 billion parameters on a single graphics processing unit. Developed by Microsoft, DeepSpeed is specifically tailored to support distributed training for extensive models, and it is constructed upon the PyTorch framework, which excels in data parallelism. Additionally, the library continuously evolves to incorporate cutting-edge advancements in deep learning, ensuring it remains at the forefront of AI technology.

DeepSeek-V4-Pro is an advanced Mixture-of-Experts language model built for high-performance reasoning, coding, and large-scale AI applications. With 1.6 trillion total parameters and 49 billion activated parameters, it delivers strong capabilities while maintaining computational efficiency. The model supports a massive context window of up to one million tokens, making it ideal for handling long documents and complex workflows. Its hybrid attention architecture improves efficiency by reducing computational overhead while maintaining accuracy. Trained on more than 32 trillion tokens, DeepSeek-V4-Pro demonstrates strong performance across knowledge, reasoning, and coding benchmarks. It includes advanced training techniques such as improved optimization and enhanced signal propagation for better stability. The model offers multiple reasoning modes, allowing users to choose between faster responses or deeper analytical thinking. It is designed to support agentic workflows and complex multi-step problem solving. As an open-source model, it provides flexibility for developers and organizations to customize and deploy at scale. Overall, DeepSeek-V4-Pro delivers a balance of performance, efficiency, and scalability for demanding AI applications.

NVIDIA NeMo Megatron serves as a comprehensive framework designed for the training and deployment of large language models (LLMs) that can range from billions to trillions of parameters. As a integral component of the NVIDIA AI platform, it provides a streamlined, efficient, and cost-effective solution in a containerized format for constructing and deploying LLMs. Tailored for enterprise application development, the framework leverages cutting-edge technologies stemming from NVIDIA research and offers a complete workflow that automates distributed data processing, facilitates the training of large-scale custom models like GPT-3, T5, and multilingual T5 (mT5), and supports model deployment for large-scale inference. The process of utilizing LLMs becomes straightforward with the availability of validated recipes and predefined configurations that streamline both training and inference. Additionally, the hyperparameter optimization tool simplifies the customization of models by automatically exploring the optimal hyperparameter configurations, enhancing performance for training and inference across various distributed GPU cluster setups. This approach not only saves time but also ensures that users can achieve superior results with minimal effort.

NVIDIA NeMo LLM offers a streamlined approach to personalizing and utilizing large language models that are built on a variety of frameworks. Developers are empowered to implement enterprise AI solutions utilizing NeMo LLM across both private and public cloud environments. They can access Megatron 530B, which is among the largest language models available, via the cloud API or through the LLM service for hands-on experimentation. Users can tailor their selections from a range of NVIDIA or community-supported models that align with their AI application needs. By utilizing prompt learning techniques, they can enhance the quality of responses in just minutes to hours by supplying targeted context for particular use cases. Moreover, the NeMo LLM Service and the cloud API allow users to harness the capabilities of NVIDIA Megatron 530B, ensuring they have access to cutting-edge language processing technology. Additionally, the platform supports models specifically designed for drug discovery, available through both the cloud API and the NVIDIA BioNeMo framework, further expanding the potential applications of this innovative service.

VideoPoet is an innovative modeling technique that transforms any autoregressive language model or large language model (LLM) into an effective video generator. It comprises several straightforward components. An autoregressive language model is trained across multiple modalities—video, image, audio, and text—to predict the subsequent video or audio token in a sequence. The training framework for the LLM incorporates a range of multimodal generative learning objectives, such as text-to-video, text-to-image, image-to-video, video frame continuation, inpainting and outpainting of videos, video stylization, and video-to-audio conversion. Additionally, these tasks can be combined to enhance zero-shot capabilities. This straightforward approach demonstrates that language models are capable of generating and editing videos with impressive temporal coherence, showcasing the potential for advanced multimedia applications. As a result, VideoPoet opens up exciting possibilities for creative expression and automated content creation.

ALBERT is a self-supervised Transformer architecture that undergoes pretraining on a vast dataset of English text, eliminating the need for manual annotations by employing an automated method to create inputs and corresponding labels from unprocessed text. This model is designed with two primary training objectives in mind. The first objective, known as Masked Language Modeling (MLM), involves randomly obscuring 15% of the words in a given sentence and challenging the model to accurately predict those masked words. This approach sets it apart from recurrent neural networks (RNNs) and autoregressive models such as GPT, as it enables ALBERT to capture bidirectional representations of sentences. The second training objective is Sentence Ordering Prediction (SOP), which focuses on the task of determining the correct sequence of two adjacent text segments during the pretraining phase. By incorporating these dual objectives, ALBERT enhances its understanding of language structure and contextual relationships. This innovative design contributes to its effectiveness in various natural language processing tasks.

BLOOM is a sophisticated autoregressive language model designed to extend text based on given prompts, leveraging extensive text data and significant computational power. This capability allows it to generate coherent and contextually relevant content in 46 different languages, along with 13 programming languages, often making it difficult to differentiate its output from that of a human author. Furthermore, BLOOM's versatility enables it to tackle various text-related challenges, even those it has not been specifically trained on, by interpreting them as tasks of text generation. Its adaptability makes it a valuable tool for a range of applications across multiple domains.

Alpa is designed to simplify the process of automating extensive distributed training and serving with minimal coding effort. Originally created by a team at Sky Lab, UC Berkeley, it employs several advanced techniques documented in a paper presented at OSDI'2022. The Alpa community continues to expand, welcoming new contributors from Google. A language model serves as a probability distribution over sequences of words, allowing it to foresee the next word based on the context of preceding words. This capability proves valuable for various AI applications, including email auto-completion and chatbot functionalities. For further insights, one can visit the Wikipedia page dedicated to language models. Among these models, GPT-3 stands out as a remarkably large language model, boasting 175 billion parameters and utilizing deep learning to generate text that closely resembles human writing. Many researchers and media outlets have characterized GPT-3 as "one of the most interesting and significant AI systems ever developed," and its influence continues to grow as it becomes integral to cutting-edge NLP research and applications. Additionally, its implementation has sparked discussions about the future of AI-driven communication tools.

Mercury, the groundbreaking creation from Inception Labs, represents the first large language model at a commercial scale that utilizes diffusion technology, achieving a remarkable tenfold increase in processing speed while also lowering costs in comparison to standard autoregressive models. Designed for exceptional performance in reasoning, coding, and the generation of structured text, Mercury can handle over 1000 tokens per second when operating on NVIDIA H100 GPUs, positioning it as one of the most rapid LLMs on the market. In contrast to traditional models that produce text sequentially, Mercury enhances its responses through a coarse-to-fine diffusion strategy, which boosts precision and minimizes instances of hallucination. Additionally, with the inclusion of Mercury Coder, a tailored coding module, developers are empowered to take advantage of advanced AI-assisted code generation that boasts remarkable speed and effectiveness. This innovative approach not only transforms coding practices but also sets a new benchmark for the capabilities of AI in various applications.

Sky-T1-32B-Preview is an innovative open-source reasoning model crafted by the NovaSky team at UC Berkeley's Sky Computing Lab. It delivers performance comparable to proprietary models such as o1-preview on various reasoning and coding assessments, while being developed at a cost of less than $450, highlighting the potential for budget-friendly, advanced reasoning abilities. Fine-tuned from Qwen2.5-32B-Instruct, the model utilized a meticulously curated dataset comprising 17,000 examples spanning multiple fields, such as mathematics and programming. The entire training process was completed in just 19 hours using eight H100 GPUs with DeepSpeed Zero-3 offloading technology. Every component of this initiative—including the data, code, and model weights—is entirely open-source, allowing both academic and open-source communities to not only replicate but also improve upon the model's capabilities. This accessibility fosters collaboration and innovation in the realm of artificial intelligence research and development.

Qwen-7B is the 7-billion parameter iteration of Alibaba Cloud's Qwen language model series, also known as Tongyi Qianwen. This large language model utilizes a Transformer architecture and has been pretrained on an extensive dataset comprising web texts, books, code, and more. Furthermore, we introduced Qwen-7B-Chat, an AI assistant that builds upon the pretrained Qwen-7B model and incorporates advanced alignment techniques. The Qwen-7B series boasts several notable features: It has been trained on a premium dataset, with over 2.2 trillion tokens sourced from a self-assembled collection of high-quality texts and codes across various domains, encompassing both general and specialized knowledge. Additionally, our model demonstrates exceptional performance, surpassing competitors of similar size on numerous benchmark datasets that assess capabilities in natural language understanding, mathematics, and coding tasks. This positions Qwen-7B as a leading choice in the realm of AI language models. Overall, its sophisticated training and robust design contribute to its impressive versatility and effectiveness.

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.

Amazon SageMaker Model Training streamlines the process of training and fine-tuning machine learning (ML) models at scale, significantly cutting down both time and costs while eliminating the need for infrastructure management. Users can leverage top-tier ML compute infrastructure, benefiting from SageMaker’s capability to seamlessly scale from a single GPU to thousands, adapting to demand as necessary. The pay-as-you-go model enables more effective management of training expenses, making it easier to keep costs in check. To accelerate the training of deep learning models, SageMaker’s distributed training libraries can divide extensive models and datasets across multiple AWS GPU instances, while also supporting third-party libraries like DeepSpeed, Horovod, or Megatron for added flexibility. Additionally, you can efficiently allocate system resources by choosing from a diverse range of GPUs and CPUs, including the powerful P4d.24xl instances, which are currently the fastest cloud training options available. With just one click, you can specify data locations and the desired SageMaker instances, simplifying the entire setup process for users. This user-friendly approach makes it accessible for both newcomers and experienced data scientists to maximize their ML training capabilities.

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.

PygmalionAI is a vibrant community focused on the development of open-source initiatives utilizing EleutherAI's GPT-J 6B and Meta's LLaMA models. Essentially, Pygmalion specializes in crafting AI tailored for engaging conversations and roleplaying. The actively maintained Pygmalion AI model currently features the 7B variant, derived from Meta AI's LLaMA model. Requiring a mere 18GB (or even less) of VRAM, Pygmalion demonstrates superior chat functionality compared to significantly larger language models, all while utilizing relatively limited resources. Our meticulously assembled dataset, rich in high-quality roleplaying content, guarantees that your AI companion will be the perfect partner for roleplaying scenarios. Both the model weights and the training code are entirely open-source, allowing you the freedom to modify and redistribute them for any purpose you desire. Generally, language models, such as Pygmalion, operate on GPUs, as they require swift memory access and substantial processing power to generate coherent text efficiently. As a result, users can expect a smooth and responsive interaction experience when employing Pygmalion's capabilities.

Pre-trained language models have made significant strides, achieving top-tier performance across multiple Natural Language Processing (NLP) applications. The impressive capabilities of GPT-3 highlight how increasing the scale of these models can unlock their vast potential. Recently, a comprehensive framework known as ERNIE 3.0 was introduced to pre-train large-scale models enriched with knowledge, culminating in a model boasting 10 billion parameters. This iteration of ERNIE 3.0 has surpassed the performance of existing leading models in a variety of NLP tasks. To further assess the effects of scaling, we have developed an even larger model called ERNIE 3.0 Titan, which consists of up to 260 billion parameters and is built on the PaddlePaddle platform. Additionally, we have implemented a self-supervised adversarial loss alongside a controllable language modeling loss, enabling ERNIE 3.0 Titan to produce texts that are both reliable and modifiable, thus pushing the boundaries of what these models can achieve. This approach not only enhances the model's capabilities but also opens new avenues for research in text generation and control.

Inception Labs is at the forefront of advancing artificial intelligence through the development of diffusion-based large language models (dLLMs), which represent a significant innovation in the field by achieving performance that is ten times faster and costs that are five to ten times lower than conventional autoregressive models. Drawing inspiration from the achievements of diffusion techniques in generating images and videos, Inception's dLLMs offer improved reasoning abilities, error correction features, and support for multimodal inputs, which collectively enhance the generation of structured and precise text. This innovative approach not only boosts efficiency but also elevates the control users have over AI outputs. With its wide-ranging applications in enterprise solutions, academic research, and content creation, Inception Labs is redefining the benchmarks for speed and effectiveness in AI-powered processes. The transformative potential of these advancements promises to reshape various industries by optimizing workflows and enhancing productivity.

GPT-4, or Generative Pre-trained Transformer 4, is a highly advanced unsupervised language model that is anticipated for release by OpenAI. As the successor to GPT-3, it belongs to the GPT-n series of natural language processing models and was developed using an extensive dataset comprising 45TB of text, enabling it to generate and comprehend text in a manner akin to human communication. Distinct from many conventional NLP models, GPT-4 operates without the need for additional training data tailored to specific tasks. It is capable of generating text or responding to inquiries by utilizing only the context it creates internally. Demonstrating remarkable versatility, GPT-4 can adeptly tackle a diverse array of tasks such as translation, summarization, question answering, sentiment analysis, and more, all without any dedicated task-specific training. This ability to perform such varied functions further highlights its potential impact on the field of artificial intelligence and natural language processing.

Stable LM represents a significant advancement in the field of language models by leveraging our previous experience with open-source initiatives, particularly in collaboration with EleutherAI, a nonprofit research organization. This journey includes the development of notable models such as GPT-J, GPT-NeoX, and the Pythia suite, all of which were trained on The Pile open-source dataset, while many contemporary open-source models like Cerebras-GPT and Dolly-2 have drawn inspiration from this foundational work. Unlike its predecessors, Stable LM is trained on an innovative dataset that is three times the size of The Pile, encompassing a staggering 1.5 trillion tokens. We plan to share more information about this dataset in the near future. The extensive nature of this dataset enables Stable LM to excel remarkably in both conversational and coding scenarios, despite its relatively modest size of 3 to 7 billion parameters when compared to larger models like GPT-3, which boasts 175 billion parameters. Designed for versatility, Stable LM 3B is a streamlined model that can efficiently function on portable devices such as laptops and handheld gadgets, making us enthusiastic about its practical applications and mobility. Overall, the development of Stable LM marks a pivotal step towards creating more efficient and accessible language models for a wider audience.

Gemini Flash represents a cutting-edge large language model developed by Google, specifically engineered for rapid, efficient language processing activities. As a part of the Gemini lineup from Google DeepMind, it is designed to deliver instantaneous responses and effectively manage extensive applications, proving to be exceptionally suited for dynamic AI-driven interactions like customer service, virtual assistants, and real-time chat systems. In addition to its impressive speed, Gemini Flash maintains a high standard of quality; it utilizes advanced neural architectures that guarantee responses are contextually appropriate, coherent, and accurate. Google has also integrated stringent ethical guidelines and responsible AI methodologies into Gemini Flash, providing it with safeguards to address and reduce biased outputs, thereby ensuring compliance with Google’s principles for secure and inclusive AI. With the capabilities of Gemini Flash, businesses and developers are empowered to implement agile, intelligent language solutions that can satisfy the requirements of rapidly evolving environments. This innovative model marks a significant step forward in the quest for sophisticated AI technologies that respect ethical considerations while enhancing user experience.

The Megatron-Turing Natural Language Generation model (MT-NLG) stands out as the largest and most advanced monolithic transformer model for the English language, boasting an impressive 530 billion parameters. This 105-layer transformer architecture significantly enhances the capabilities of previous leading models, particularly in zero-shot, one-shot, and few-shot scenarios. It exhibits exceptional precision across a wide range of natural language processing tasks, including completion prediction, reading comprehension, commonsense reasoning, natural language inference, and word sense disambiguation. To foster further research on this groundbreaking English language model and to allow users to explore and utilize its potential in various language applications, NVIDIA has introduced an Early Access program for its managed API service dedicated to the MT-NLG model. This initiative aims to facilitate experimentation and innovation in the field of natural language processing.

Utilize sophisticated coding and language models across a diverse range of applications. Harness the power of expansive generative AI models that possess an intricate grasp of both language and code, paving the way for enhanced reasoning and comprehension skills essential for developing innovative applications. These advanced models can be applied to multiple scenarios, including writing support, automatic code creation, and data reasoning. Moreover, ensure responsible AI practices by implementing measures to detect and mitigate potential misuse, all while benefiting from enterprise-level security features offered by Azure. With access to generative models pretrained on vast datasets comprising trillions of words, you can explore new possibilities in language processing, code analysis, reasoning, inferencing, and comprehension. Further personalize these generative models by using labeled datasets tailored to your unique needs through an easy-to-use REST API. Additionally, you can optimize your model's performance by fine-tuning hyperparameters for improved output accuracy. The few-shot learning functionality allows you to provide sample inputs to the API, resulting in more pertinent and context-aware outcomes. This flexibility enhances your ability to meet specific application demands effectively.

Dolly is an economical large language model that surprisingly demonstrates a notable level of instruction-following abilities similar to those seen in ChatGPT. While the Alpaca team's research revealed that cutting-edge models could be encouraged to excel in high-quality instruction adherence, our findings indicate that even older open-source models with earlier architectures can display remarkable behaviors when fine-tuned on a modest set of instructional training data. By utilizing an existing open-source model with 6 billion parameters from EleutherAI, Dolly has been slightly adjusted to enhance its ability to follow instructions, showcasing skills like brainstorming and generating text that were absent in its original form. This approach not only highlights the potential of older models but also opens new avenues for leveraging existing technologies in innovative ways.

NVIDIA's Nemotron 3 represents a collection of open large language models crafted to drive advanced reasoning, conversational AI, and autonomous AI agents. This series consists of three distinct models tailored for varying scales of AI workloads, all while ensuring remarkable efficiency and precision. Emphasizing "agentic AI" features, these models are capable of executing multi-step reasoning, collaborating with tools, and functioning as integral parts of multi-agent systems utilized across automation, research, and enterprise sectors. The underlying architecture employs a hybrid mixture-of-experts (MoE) approach paired with transformer techniques, enabling the activation of only specific parameter subsets for each task, thereby enhancing performance and minimizing computational expenses. Designed to excel in reasoning, dialogue, and strategic planning, the Nemotron 3 models are optimized for high throughput, making them suitable for extensive deployment across diverse applications. Additionally, their innovative architecture allows for greater adaptability and scalability, ensuring they meet the evolving demands of modern AI challenges.

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.

SubQ is an advanced large language model created by Subquadratic to handle complex long-context reasoning tasks. It supports up to 12 million tokens in a single input, making it capable of analyzing entire repositories, extended conversation histories, and large datasets without losing context. The model is built on a sub-quadratic sparse-attention architecture that focuses computational resources on the most relevant data relationships. This design significantly reduces processing requirements compared to traditional transformer models while maintaining strong performance. SubQ is particularly useful for software engineering, coding workflows, and long-context retrieval tasks. It enables developers and teams to process large amounts of information in a single operation instead of splitting tasks into smaller parts. The model offers fast processing speeds and operates at a fraction of the cost of many competing solutions. It is available through API access, allowing integration into enterprise systems and developer tools. SubQ can also be used as a layer within coding agents to improve code exploration and analysis. Its compatibility with existing development environments makes it easier to adopt. With its efficient architecture and large context window, it helps teams work with complex data more effectively.

Baichuan-13B is an advanced large-scale language model developed by Baichuan Intelligent, featuring 13 billion parameters and available for open-source and commercial use, building upon its predecessor Baichuan-7B. This model has set new records for performance among similarly sized models on esteemed Chinese and English evaluation metrics. The release includes two distinct pre-training variations: Baichuan-13B-Base and Baichuan-13B-Chat. By significantly increasing the parameter count to 13 billion, Baichuan-13B enhances its capabilities, training on 1.4 trillion tokens from a high-quality dataset, which surpasses LLaMA-13B's training data by 40%. It currently holds the distinction of being the model with the most extensive training data in the 13B category, providing robust support for both Chinese and English languages, utilizing ALiBi positional encoding, and accommodating a context window of 4096 tokens for improved comprehension and generation. This makes it a powerful tool for a variety of applications in natural language processing.

Training cutting-edge language models presents significant challenges; it demands vast computational resources, intricate distributed computing strategies, and substantial machine learning knowledge. Consequently, only a limited number of organizations embark on the journey of developing large language models (LLMs) from the ground up. Furthermore, many of those with the necessary capabilities and knowledge have begun to restrict access to their findings, indicating a notable shift from practices observed just a few months ago. At Cerebras, we are committed to promoting open access to state-of-the-art models. Therefore, we are excited to share with the open-source community the launch of Cerebras-GPT, which consists of a series of seven GPT models with parameter counts ranging from 111 million to 13 billion. Utilizing the Chinchilla formula for training, these models deliver exceptional accuracy while optimizing for computational efficiency. Notably, Cerebras-GPT boasts quicker training durations, reduced costs, and lower energy consumption compared to any publicly accessible model currently available. By releasing these models, we hope to inspire further innovation and collaboration in the field of machine learning.

NVIDIA has created the Nemotron family of open-source models aimed at producing synthetic data specifically for training large language models (LLMs) intended for commercial use. Among these, the Nemotron-4 340B model stands out as a key innovation, providing developers with a robust resource to generate superior quality data while also allowing for the filtering of this data according to multiple attributes through a reward model. This advancement not only enhances data generation capabilities but also streamlines the process of training LLMs, making it more efficient and tailored to specific needs.

Inflection AI is an innovative research and development company in the realm of artificial intelligence, dedicated to crafting sophisticated AI systems that facilitate more natural and intuitive interactions with humans. Established in 2022 by notable entrepreneurs including Mustafa Suleyman, who co-founded DeepMind, and Reid Hoffman, a co-founder of LinkedIn, the company aims to democratize access to powerful AI while ensuring it aligns closely with human values. Inflection AI concentrates on developing extensive language models that improve communication between humans and AI, with the intention of revolutionizing various sectors, including customer support and personal productivity, through the implementation of intelligent, responsive, and ethically conceived AI systems. With a strong emphasis on safety, transparency, and user empowerment, the company is committed to ensuring that its advancements have a constructive impact on society, all while actively mitigating the potential risks linked to AI technologies. Moreover, Inflection AI aspires to pave the way for future innovations that prioritize both utility and ethical considerations, reinforcing its role as a leader in the AI landscape.

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.

Qwen3.5-Plus is an advanced multimodal foundation model engineered to deliver efficient large-context reasoning across text, image, and video inputs. Powered by a hybrid architecture that merges linear attention mechanisms with a sparse mixture-of-experts framework, the model achieves state-of-the-art performance while reducing computational overhead. It supports deep thinking mode, enabling extended reasoning chains of up to 80K tokens and total context windows of up to 1 million tokens. Developers can leverage features such as structured output generation, function calling, web search, and integrated code interpretation to build intelligent agent workflows. The model is optimized for high throughput, supporting large token-per-minute limits and robust rate limits for enterprise-scale applications. Qwen3.5-Plus also includes explicit caching options to reduce costs during repeated inference tasks. With tiered pricing based on input and output tokens, organizations can scale usage predictably. OpenAI-compatible API endpoints make integration straightforward across existing AI stacks and developer tools. Designed for demanding applications, Qwen3.5-Plus excels in long-document analysis, multimodal reasoning, and advanced AI agent development.

Gemini Diffusion represents our cutting-edge research initiative aimed at redefining the concept of diffusion in the realm of language and text generation. Today, large language models serve as the backbone of generative AI technology. By employing a diffusion technique, we are pioneering a new type of language model that enhances user control, fosters creativity, and accelerates the text generation process. Unlike traditional models that predict text in a straightforward manner, diffusion models take a unique approach by generating outputs through a gradual refinement of noise. This iterative process enables them to quickly converge on solutions and make real-time corrections during generation. As a result, they demonstrate superior capabilities in tasks such as editing, particularly in mathematics and coding scenarios. Furthermore, by generating entire blocks of tokens simultaneously, they provide more coherent responses to user prompts compared to autoregressive models. Remarkably, the performance of Gemini Diffusion on external benchmarks rivals that of much larger models, while also delivering enhanced speed, making it a noteworthy advancement in the field. This innovation not only streamlines the generation process but also opens new avenues for creative expression in language-based tasks.

MiniMax M2 is an open-source foundational model tailored for agent-driven applications and coding tasks, achieving an innovative equilibrium of efficiency, velocity, and affordability. It shines in comprehensive development environments, adeptly managing programming tasks, invoking tools, and executing intricate, multi-step processes, complete with features like Python integration, while offering impressive inference speeds of approximately 100 tokens per second and competitive API pricing at around 8% of similar proprietary models. The model includes a "Lightning Mode" designed for rapid, streamlined agent operations, alongside a "Pro Mode" aimed at thorough full-stack development, report creation, and the orchestration of web-based tools; its weights are entirely open source, allowing for local deployment via vLLM or SGLang. MiniMax M2 stands out as a model ready for production use, empowering agents to autonomously perform tasks such as data analysis, software development, tool orchestration, and implementing large-scale, multi-step logic across real organizational contexts. With its advanced capabilities, this model is poised to revolutionize the way developers approach complex programming challenges.

Qwen2.5-Max is an advanced Mixture-of-Experts (MoE) model created by the Qwen team, which has been pretrained on an extensive dataset of over 20 trillion tokens and subsequently enhanced through methods like Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF). Its performance in evaluations surpasses that of models such as DeepSeek V3 across various benchmarks, including Arena-Hard, LiveBench, LiveCodeBench, and GPQA-Diamond, while also achieving strong results in other tests like MMLU-Pro. This model is available through an API on Alibaba Cloud, allowing users to easily integrate it into their applications, and it can also be interacted with on Qwen Chat for a hands-on experience. With its superior capabilities, Qwen2.5-Max represents a significant advancement in AI model technology.

DeepScaleR is a sophisticated language model comprising 1.5 billion parameters, refined from DeepSeek-R1-Distilled-Qwen-1.5B through the use of distributed reinforcement learning combined with an innovative strategy that incrementally expands its context window from 8,000 to 24,000 tokens during the training process. This model was developed using approximately 40,000 meticulously selected mathematical problems sourced from high-level competition datasets, including AIME (1984–2023), AMC (pre-2023), Omni-MATH, and STILL. Achieving an impressive 43.1% accuracy on the AIME 2024 exam, DeepScaleR demonstrates a significant enhancement of around 14.3 percentage points compared to its base model, and it even outperforms the proprietary O1-Preview model, which is considerably larger. Additionally, it excels on a variety of mathematical benchmarks such as MATH-500, AMC 2023, Minerva Math, and OlympiadBench, indicating that smaller, optimized models fine-tuned with reinforcement learning can rival or surpass the capabilities of larger models in complex reasoning tasks. This advancement underscores the potential of efficient modeling approaches in the realm of mathematical problem-solving.

GPT-5.4 nano is a compact and cost-efficient AI model designed for handling lightweight, high-frequency tasks at scale. It is optimized for operations such as classification, data extraction, ranking, and simple coding assistance. The model delivers fast response times, making it suitable for applications where low latency is critical. Compared to earlier nano models, GPT-5.4 nano offers improved performance while maintaining minimal computational cost. It supports key features such as tool usage and structured output generation, allowing it to integrate easily into automated systems. The model is often used as a subagent within larger AI workflows, handling repetitive or supporting tasks efficiently. This approach allows more complex models to focus on higher-level reasoning and decision-making. GPT-5.4 nano is particularly useful in environments that require processing large volumes of requests quickly. Its efficiency makes it ideal for cost-sensitive applications and scalable deployments. Overall, it provides a reliable and fast solution for simple AI-driven tasks.

GLM-OCR is an advanced multimodal optical character recognition system and an open-source framework that excels in delivering precise, efficient, and thorough document comprehension by integrating textual and visual elements within a cohesive encoder-decoder design inspired by the GLM-V series. This model features a visual encoder that has been pre-trained on extensive image-text datasets alongside a streamlined cross-modal connector that channels information into a GLM-0.5B language decoder. It offers capabilities for layout detection, simultaneous recognition of various regions, and structured outputs for diverse content types, including text, tables, formulas, and intricate real-world document formats. Furthermore, it employs Multi-Token Prediction (MTP) loss and robust full-task reinforcement learning techniques to enhance training efficiency, boost recognition accuracy, and improve generalization across various tasks, leading to remarkable performance on significant document understanding challenges. This innovative approach not only sets new benchmarks but also opens up possibilities for further advancements in the field of document analysis.

Qwen3.6-Max-Preview represents an advanced frontier language model aimed at enhancing intelligence, following instructions, and improving real-world agent functionalities within the Qwen ecosystem. This preview builds upon the Qwen3 series, showcasing enhanced world knowledge, refined alignment with instructions, and notable advancements in coding performance for agents, which allows the model to adeptly manage intricate, multi-step tasks and software engineering processes. It is meticulously designed for scenarios requiring advanced reasoning and execution, where the model goes beyond merely generating responses to actively interacting with tools, processing lengthy contexts, and facilitating structured problem-solving in various fields such as coding, research, and enterprise operations. The architecture continues to embody the Qwen commitment to developing large-scale, high-efficiency models that can effectively manage extensive context windows while providing reliable performance across multilingual and knowledge-intensive projects. Moreover, its capabilities promise to significantly enhance productivity and innovation in diverse applications.

Qwen3 is a state-of-the-art large language model designed to revolutionize the way we interact with AI. Featuring both thinking and non-thinking modes, Qwen3 allows users to customize its response style, ensuring optimal performance for both complex reasoning tasks and quick inquiries. With the ability to support 119 languages, the model is suitable for international projects. The model's hybrid training approach, which involves over 36 trillion tokens, ensures accuracy across a variety of disciplines, from coding to STEM problems. Its integration with platforms such as Hugging Face, ModelScope, and Kaggle allows for easy adoption in both research and production environments. By enhancing multilingual support and incorporating advanced AI techniques, Qwen3 is designed to push the boundaries of AI-driven applications.