Alternatives to BitNet

ChatGLM-6B is a bilingual dialogue model that supports both Chinese and English, built on the General Language Model (GLM) framework and features 6.2 billion parameters. Thanks to model quantization techniques, it can be easily run on standard consumer graphics cards, requiring only 6GB of video memory at the INT4 quantization level. This model employs methodologies akin to those found in ChatGPT but is specifically tailored to enhance Chinese question-and-answer interactions and dialogue. Following extensive training with approximately 1 trillion identifiers in both languages, along with additional supervision, fine-tuning, self-assistance through feedback, and reinforcement learning from human input, ChatGLM-6B has demonstrated an impressive capability to produce responses that resonate well with human users. Its adaptability and performance make it a valuable tool for bilingual communication.

Kimi K2 Thinking is a sophisticated open-source reasoning model created by Moonshot AI, specifically tailored for intricate, multi-step workflows where it effectively combines chain-of-thought reasoning with tool utilization across numerous sequential tasks. Employing a cutting-edge mixture-of-experts architecture, the model encompasses a staggering total of 1 trillion parameters, although only around 32 billion parameters are utilized during each inference, which enhances efficiency while retaining significant capability. It boasts a context window that can accommodate up to 256,000 tokens, allowing it to process exceptionally long inputs and reasoning sequences without sacrificing coherence. Additionally, it features native INT4 quantization, which significantly cuts down inference latency and memory consumption without compromising performance. Designed with agentic workflows in mind, Kimi K2 Thinking is capable of autonomously invoking external tools, orchestrating sequential logic steps—often involving around 200-300 tool calls in a single chain—and ensuring consistent reasoning throughout the process. Its robust architecture makes it an ideal solution for complex reasoning tasks that require both depth and efficiency.

Recent breakthroughs in natural language processing, comprehension, and generation have been greatly influenced by the development of large language models. This research presents a system that employs Ascend 910 AI processors and the MindSpore framework to train a language model exceeding one trillion parameters, specifically 1.085 trillion, referred to as PanGu-{\Sigma}. This model enhances the groundwork established by PanGu-{\alpha} by converting the conventional dense Transformer model into a sparse format through a method known as Random Routed Experts (RRE). Utilizing a substantial dataset of 329 billion tokens, the model was effectively trained using a strategy called Expert Computation and Storage Separation (ECSS), which resulted in a remarkable 6.3-fold improvement in training throughput through the use of heterogeneous computing. Through various experiments, it was found that PanGu-{\Sigma} achieves a new benchmark in zero-shot learning across multiple downstream tasks in Chinese NLP, showcasing its potential in advancing the field. This advancement signifies a major leap forward in the capabilities of language models, illustrating the impact of innovative training techniques and architectural modifications.

Reka Flash 3 is a cutting-edge multimodal AI model with 21 billion parameters, crafted by Reka AI to perform exceptionally well in tasks such as general conversation, coding, following instructions, and executing functions. This model adeptly handles and analyzes a myriad of inputs, including text, images, video, and audio, providing a versatile and compact solution for a wide range of applications. Built from the ground up, Reka Flash 3 was trained on a rich array of datasets, encompassing both publicly available and synthetic information, and it underwent a meticulous instruction tuning process with high-quality selected data to fine-tune its capabilities. The final phase of its training involved employing reinforcement learning techniques, specifically using the REINFORCE Leave One-Out (RLOO) method, which combined both model-based and rule-based rewards to significantly improve its reasoning skills. With an impressive context length of 32,000 tokens, Reka Flash 3 competes effectively with proprietary models like OpenAI's o1-mini, making it an excellent choice for applications requiring low latency or on-device processing. The model operates at full precision with a memory requirement of 39GB (fp16), although it can be efficiently reduced to just 11GB through the use of 4-bit quantization, demonstrating its adaptability for various deployment scenarios. Overall, Reka Flash 3 represents a significant advancement in multimodal AI technology, capable of meeting diverse user needs across multiple platforms.

DeepSeek-V2 is a cutting-edge Mixture-of-Experts (MoE) language model developed by DeepSeek-AI, noted for its cost-effective training and high-efficiency inference features. It boasts an impressive total of 236 billion parameters, with only 21 billion active for each token, and is capable of handling a context length of up to 128K tokens. The model utilizes advanced architectures such as Multi-head Latent Attention (MLA) to optimize inference by minimizing the Key-Value (KV) cache and DeepSeekMoE to enable economical training through sparse computations. Compared to its predecessor, DeepSeek 67B, this model shows remarkable improvements, achieving a 42.5% reduction in training expenses, a 93.3% decrease in KV cache size, and a 5.76-fold increase in generation throughput. Trained on an extensive corpus of 8.1 trillion tokens, DeepSeek-V2 demonstrates exceptional capabilities in language comprehension, programming, and reasoning tasks, positioning it as one of the leading open-source models available today. Its innovative approach not only elevates its performance but also sets new benchmarks within the field of artificial intelligence.

Kimi K2 represents a cutting-edge series of open-source large language models utilizing a mixture-of-experts (MoE) architecture, with a staggering 1 trillion parameters in total and 32 billion activated parameters tailored for optimized task execution. Utilizing the Muon optimizer, it has been trained on a substantial dataset of over 15.5 trillion tokens, with its performance enhanced by MuonClip’s attention-logit clamping mechanism, resulting in remarkable capabilities in areas such as advanced knowledge comprehension, logical reasoning, mathematics, programming, and various agentic operations. Moonshot AI offers two distinct versions: Kimi-K2-Base, designed for research-level fine-tuning, and Kimi-K2-Instruct, which is pre-trained for immediate applications in chat and tool interactions, facilitating both customized development and seamless integration of agentic features. Comparative benchmarks indicate that Kimi K2 surpasses other leading open-source models and competes effectively with top proprietary systems, particularly excelling in coding and intricate task analysis. Furthermore, it boasts a generous context length of 128 K tokens, compatibility with tool-calling APIs, and support for industry-standard inference engines, making it a versatile option for various applications. The innovative design and features of Kimi K2 position it as a significant advancement in the field of artificial intelligence language processing.

Olmo 3 represents a comprehensive family of open models featuring variations with 7 billion and 32 billion parameters, offering exceptional capabilities in base performance, reasoning, instruction, and reinforcement learning, while also providing transparency throughout the model development process, which includes access to raw training datasets, intermediate checkpoints, training scripts, extended context support (with a window of 65,536 tokens), and provenance tools. The foundation of these models is built upon the Dolma 3 dataset, which comprises approximately 9 trillion tokens and utilizes a careful blend of web content, scientific papers, programming code, and lengthy documents; this thorough pre-training, mid-training, and long-context approach culminates in base models that undergo post-training enhancements through supervised fine-tuning, preference optimization, and reinforcement learning with accountable rewards, resulting in the creation of the Think and Instruct variants. Notably, the 32 billion Think model has been recognized as the most powerful fully open reasoning model to date, demonstrating performance that closely rivals that of proprietary counterparts in areas such as mathematics, programming, and intricate reasoning tasks, thereby marking a significant advancement in open model development. This innovation underscores the potential for open-source models to compete with traditional, closed systems in various complex applications.

Llama 4 Scout is an advanced multimodal AI model with 17 billion active parameters, offering industry-leading performance with a 10 million token context length. This enables it to handle complex tasks like multi-document summarization and detailed code reasoning with impressive accuracy. Scout surpasses previous Llama models in both text and image understanding, making it an excellent choice for applications that require a combination of language processing and image analysis. Its powerful capabilities in long-context tasks and image-grounding applications set it apart from other models in its class, providing superior results for a wide range of industries.

GigaChat 3 Ultra redefines open-source scale by delivering a 702B-parameter frontier model purpose-built for Russian and multilingual understanding. Designed with a modern MoE architecture, it achieves the reasoning strength of giant dense models while using only a fraction of active parameters per generation step. Its massive 14T-token training corpus includes natural human text, curated multilingual sources, extensive STEM materials, and billions of high-quality synthetic examples crafted to boost logic, math, and programming skills. This model is not a derivative or retrained foreign LLM—it is a ground-up build engineered to capture cultural nuance, linguistic accuracy, and reliable long-context performance. GigaChat 3 Ultra integrates seamlessly with open-source tooling like vLLM, sglang, DeepSeek-class architectures, and HuggingFace-based training stacks. It supports advanced capabilities including a code interpreter, improved chat template, memory system, contextual search reformulation, and 128K context windows. Benchmarking shows clear improvements over previous GigaChat generations and competitive results against global leaders in coding, reasoning, and cross-domain tasks. Overall, GigaChat 3 Ultra empowers teams to explore frontier-scale AI without sacrificing transparency, customizability, or ecosystem compatibility.

The TinyLlama initiative seeks to pretrain a Llama model with 1.1 billion parameters using a dataset of 3 trillion tokens. With the right optimizations, this ambitious task can be completed in a mere 90 days, utilizing 16 A100-40G GPUs. We have maintained the same architecture and tokenizer as Llama 2, ensuring that TinyLlama is compatible with various open-source projects that are based on Llama. Additionally, the model's compact design, consisting of just 1.1 billion parameters, makes it suitable for numerous applications that require limited computational resources and memory. This versatility enables developers to integrate TinyLlama seamlessly into their existing frameworks and workflows.

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.

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.

The QwQ-32B model, created by Alibaba Cloud's Qwen team, represents a significant advancement in AI reasoning, aimed at improving problem-solving skills. Boasting 32 billion parameters, it rivals leading models such as DeepSeek's R1, which contains 671 billion parameters. This remarkable efficiency stems from its optimized use of parameters, enabling QwQ-32B to tackle complex tasks like mathematical reasoning, programming, and other problem-solving scenarios while consuming fewer resources. It can handle a context length of up to 32,000 tokens, making it adept at managing large volumes of input data. Notably, QwQ-32B is available through Alibaba's Qwen Chat service and is released under the Apache 2.0 license, which fosters collaboration and innovation among AI developers. With its cutting-edge features, QwQ-32B is poised to make a substantial impact in the field of artificial intelligence.

Mistral AI has unveiled two cutting-edge models specifically designed for on-device computing and edge use cases, collectively referred to as "les Ministraux": Ministral 3B and Ministral 8B. These innovative models stand out due to their capabilities in knowledge retention, commonsense reasoning, function-calling, and overall efficiency, all while remaining within the sub-10B parameter range. They boast support for a context length of up to 128k, making them suitable for a diverse range of applications such as on-device translation, offline smart assistants, local analytics, and autonomous robotics. Notably, Ministral 8B incorporates an interleaved sliding-window attention mechanism, which enhances both the speed and memory efficiency of inference processes. Both models are adept at serving as intermediaries in complex multi-step workflows, skillfully managing functions like input parsing, task routing, and API interactions based on user intent, all while minimizing latency and operational costs. Benchmark results reveal that les Ministraux consistently exceed the performance of similar models across a variety of tasks, solidifying their position in the market. As of October 16, 2024, these models are now available for developers and businesses, with Ministral 8B being offered at a competitive rate of $0.1 for every million tokens utilized. This pricing structure enhances accessibility for users looking to integrate advanced AI capabilities into their solutions.

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.

Introducing Mistral NeMo, our latest and most advanced small model yet, featuring a cutting-edge 12 billion parameters and an expansive context length of 128,000 tokens, all released under the Apache 2.0 license. Developed in partnership with NVIDIA, Mistral NeMo excels in reasoning, world knowledge, and coding proficiency within its category. Its architecture adheres to industry standards, making it user-friendly and a seamless alternative for systems currently utilizing Mistral 7B. To facilitate widespread adoption among researchers and businesses, we have made available both pre-trained base and instruction-tuned checkpoints under the same Apache license. Notably, Mistral NeMo incorporates quantization awareness, allowing for FP8 inference without compromising performance. The model is also tailored for diverse global applications, adept in function calling and boasting a substantial context window. When compared to Mistral 7B, Mistral NeMo significantly outperforms in understanding and executing detailed instructions, showcasing enhanced reasoning skills and the ability to manage complex multi-turn conversations. Moreover, its design positions it as a strong contender for multi-lingual tasks, ensuring versatility across various use cases.

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.

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.

Qwen3-Max represents Alibaba's cutting-edge large language model, featuring a staggering trillion parameters aimed at enhancing capabilities in tasks that require agency, coding, reasoning, and managing lengthy contexts. This model is an evolution of the Qwen3 series, leveraging advancements in architecture, training methods, and inference techniques; it integrates both thinker and non-thinker modes, incorporates a unique “thinking budget” system, and allows for dynamic mode adjustments based on task complexity. Capable of handling exceptionally lengthy inputs, processing hundreds of thousands of tokens, it also supports tool invocation and demonstrates impressive results across various benchmarks, including coding, multi-step reasoning, and agent evaluations like Tau2-Bench. While the initial version prioritizes instruction adherence in a non-thinking mode, Alibaba is set to introduce reasoning functionalities that will facilitate autonomous agent operations in the future. In addition to its existing multilingual capabilities and extensive training on trillions of tokens, Qwen3-Max is accessible through API interfaces that align seamlessly with OpenAI-style functionalities, ensuring broad usability across applications. This comprehensive framework positions Qwen3-Max as a formidable player in the realm of advanced artificial intelligence language models.

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.

Mistral Saba is an advanced model boasting 24 billion parameters, developed using carefully selected datasets from the Middle East and South Asia. It outperforms larger models—those more than five times its size—in delivering precise and pertinent responses, all while being notably faster and more cost-effective. Additionally, it serves as an excellent foundation for creating highly specialized regional adaptations. This model can be accessed via an API and is also capable of being deployed locally to meet customers' security requirements. Similar to the recently introduced Mistral Small 3, it is lightweight enough to operate on single-GPU systems, achieving response rates exceeding 150 tokens per second. Reflecting the deep cultural connections between the Middle East and South Asia, Mistral Saba is designed to support Arabic alongside numerous Indian languages, with a particular proficiency in South Indian languages like Tamil. This diverse linguistic capability significantly boosts its adaptability for multinational applications in these closely linked regions. Furthermore, the model’s design facilitates an easier integration into various platforms, enhancing its usability across different industries.

DeepSeek-R1 is a cutting-edge open-source reasoning model created by DeepSeek, aimed at competing with OpenAI's Model o1. It is readily available through web, app, and API interfaces, showcasing its proficiency in challenging tasks such as mathematics and coding, and achieving impressive results on assessments like the American Invitational Mathematics Examination (AIME) and MATH. Utilizing a mixture of experts (MoE) architecture, this model boasts a remarkable total of 671 billion parameters, with 37 billion parameters activated for each token, which allows for both efficient and precise reasoning abilities. As a part of DeepSeek's dedication to the progression of artificial general intelligence (AGI), the model underscores the importance of open-source innovation in this field. Furthermore, its advanced capabilities may significantly impact how we approach complex problem-solving in various domains.

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.

StarCoder and StarCoderBase represent advanced Large Language Models specifically designed for code, developed using openly licensed data from GitHub, which encompasses over 80 programming languages, Git commits, GitHub issues, and Jupyter notebooks. In a manner akin to LLaMA, we constructed a model with approximately 15 billion parameters trained on a staggering 1 trillion tokens. Furthermore, we tailored the StarCoderBase model with 35 billion Python tokens, leading to the creation of what we now refer to as StarCoder. Our evaluations indicated that StarCoderBase surpasses other existing open Code LLMs when tested against popular programming benchmarks and performs on par with or even exceeds proprietary models like code-cushman-001 from OpenAI, the original Codex model that fueled early iterations of GitHub Copilot. With an impressive context length exceeding 8,000 tokens, the StarCoder models possess the capability to handle more information than any other open LLM, thus paving the way for a variety of innovative applications. This versatility is highlighted by our ability to prompt the StarCoder models through a sequence of dialogues, effectively transforming them into dynamic technical assistants that can provide support in diverse programming tasks.

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.

Google's Gemini Nano is an efficient and lightweight AI model engineered to perform exceptionally well in environments with limited resources. Specifically designed for mobile applications and edge computing, it merges Google's sophisticated AI framework with innovative optimization strategies, ensuring high-speed performance and accuracy are preserved. This compact model stands out in various applications, including voice recognition, real-time translation, natural language processing, and delivering personalized recommendations. Emphasizing both privacy and efficiency, Gemini Nano processes information locally to reduce dependence on cloud services while ensuring strong security measures are in place. Its versatility and minimal power requirements make it perfectly suited for smart devices, IoT applications, and portable AI technologies. As a result, it opens up new possibilities for developers looking to integrate advanced AI into everyday gadgets.

K2 Think represents a groundbreaking open-source advanced reasoning model that has been developed in collaboration between the Institute of Foundation Models at MBZUAI and G42. Even with its relatively modest 32 billion parameters, K2 Think achieves performance that rivals that of leading models with significantly larger parameter counts. Its strength lies in mathematical reasoning, where it has secured top rankings on prestigious benchmarks such as AIME ’24/’25, HMMT ’25, and OMNI-Math-HARD. This model is part of a wider initiative of UAE-developed open models, which includes Jais (for Arabic), NANDA (for Hindi), and SHERKALA (for Kazakh), and it builds upon the groundwork established by the K2-65B, a fully reproducible open-source foundation model released in 2024. K2 Think is crafted to be open, efficient, and adaptable, featuring a web app interface that facilitates user exploration, and its innovative approach to parameter positioning marks a significant advancement in the realm of compact architectures for high-level AI reasoning. Additionally, its development highlights a commitment to enhancing access to state-of-the-art AI technologies in various languages and domains.

Pixtral Large is an expansive multimodal model featuring 124 billion parameters, crafted by Mistral AI and enhancing their previous Mistral Large 2 framework. This model combines a 123-billion-parameter multimodal decoder with a 1-billion-parameter vision encoder, allowing it to excel in the interpretation of various content types, including documents, charts, and natural images, all while retaining superior text comprehension abilities. With the capability to manage a context window of 128,000 tokens, Pixtral Large can efficiently analyze at least 30 high-resolution images at once. It has achieved remarkable results on benchmarks like MathVista, DocVQA, and VQAv2, outpacing competitors such as GPT-4o and Gemini-1.5 Pro. Available for research and educational purposes under the Mistral Research License, it also has a Mistral Commercial License for business applications. This versatility makes Pixtral Large a valuable tool for both academic research and commercial innovations.

OLMo 2 represents a collection of completely open language models created by the Allen Institute for AI (AI2), aimed at giving researchers and developers clear access to training datasets, open-source code, reproducible training methodologies, and thorough assessments. These models are trained on an impressive volume of up to 5 trillion tokens and compete effectively with top open-weight models like Llama 3.1, particularly in English academic evaluations. A key focus of OLMo 2 is on ensuring training stability, employing strategies to mitigate loss spikes during extended training periods, and applying staged training interventions in the later stages of pretraining to mitigate weaknesses in capabilities. Additionally, the models leverage cutting-edge post-training techniques derived from AI2's Tülu 3, leading to the development of OLMo 2-Instruct models. To facilitate ongoing enhancements throughout the development process, an actionable evaluation framework known as the Open Language Modeling Evaluation System (OLMES) was created, which includes 20 benchmarks that evaluate essential capabilities. This comprehensive approach not only fosters transparency but also encourages continuous improvement in language model performance.

Introducing the next iteration of our open-source large language model, this version features model weights along with initial code for the pretrained and fine-tuned Llama language models, which span from 7 billion to 70 billion parameters. The Llama 2 pretrained models have been developed using an impressive 2 trillion tokens and offer double the context length compared to their predecessor, Llama 1. Furthermore, the fine-tuned models have been enhanced through the analysis of over 1 million human annotations. Llama 2 demonstrates superior performance against various other open-source language models across multiple external benchmarks, excelling in areas such as reasoning, coding capabilities, proficiency, and knowledge assessments. For its training, Llama 2 utilized publicly accessible online data sources, while the fine-tuned variant, Llama-2-chat, incorporates publicly available instruction datasets along with the aforementioned extensive human annotations. Our initiative enjoys strong support from a diverse array of global stakeholders who are enthusiastic about our open approach to AI, including companies that have provided valuable early feedback and are eager to collaborate using Llama 2. The excitement surrounding Llama 2 signifies a pivotal shift in how AI can be developed and utilized collectively.

GPT-5.1 Thinking represents an evolved reasoning model within the GPT-5.1 lineup, engineered to optimize "thinking time" allocation according to the complexity of prompts, allowing for quicker responses to straightforward inquiries while dedicating more resources to tackle challenging issues. In comparison to its earlier version, it demonstrates approximately double the speed on simpler tasks and takes twice as long for more complex ones. The model emphasizes clarity in its responses, minimizing the use of jargon and undefined terminology, which enhances the accessibility and comprehensibility of intricate analytical tasks. It adeptly modifies its reasoning depth, ensuring a more effective equilibrium between rapidity and thoroughness, especially when addressing technical subjects or multi-step inquiries. By fusing substantial reasoning power with enhanced clarity, GPT-5.1 Thinking emerges as an invaluable asset for handling complicated assignments, including in-depth analysis, programming, research, or technical discussions, while simultaneously decreasing unnecessary delays for routine requests. This improved efficiency not only benefits users seeking quick answers but also supports those engaged in more demanding cognitive tasks.

GLM-4.5V-Flash is a vision-language model that is open source and specifically crafted to integrate robust multimodal functionalities into a compact and easily deployable framework. It accommodates various types of inputs including images, videos, documents, and graphical user interfaces, facilitating a range of tasks such as understanding scenes, parsing charts and documents, reading screens, and analyzing multiple images. In contrast to its larger counterparts, GLM-4.5V-Flash maintains a smaller footprint while still embodying essential visual language model features such as visual reasoning, video comprehension, handling GUI tasks, and parsing complex documents. This model can be utilized within “GUI agent” workflows, allowing it to interpret screenshots or desktop captures, identify icons or UI components, and assist with both automated desktop and web tasks. While it may not achieve the performance enhancements seen in the largest models, GLM-4.5V-Flash is highly adaptable for practical multimodal applications where efficiency, reduced resource requirements, and extensive modality support are key considerations. Its design ensures that users can harness powerful functionalities without sacrificing speed or accessibility.

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.

Anthropic’s release of Claude Opus 4.5 introduces a frontier AI model that excels at coding, complex reasoning, deep research, and long-context tasks. It sets new performance records on real-world engineering benchmarks, handling multi-system debugging, ambiguous instructions, and cross-domain problem solving with greater precision than earlier versions. Testers and early customers reported that Opus 4.5 “just gets it,” offering creative reasoning strategies that even benchmarks fail to anticipate. Beyond raw capability, the model brings stronger alignment and safety, with notable advances in prompt-injection resistance and behavior consistency in high-stakes scenarios. The Claude Developer Platform also gains richer controls including effort tuning, multi-agent orchestration, and context management improvements that significantly boost efficiency. Claude Code becomes more powerful with enhanced planning abilities, multi-session desktop support, and better execution of complex development workflows. In the Claude apps, extended memory and automatic context summarization enable longer, uninterrupted conversations. Together, these upgrades showcase Opus 4.5 as a highly capable, secure, and versatile model designed for both professional workloads and everyday use.

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.

Chinchilla is an advanced language model that operates with a compute budget comparable to Gopher while having 70 billion parameters and utilizing four times the amount of data. This model consistently and significantly surpasses Gopher (280 billion parameters), as well as GPT-3 (175 billion), Jurassic-1 (178 billion), and Megatron-Turing NLG (530 billion), across a wide variety of evaluation tasks. Additionally, Chinchilla's design allows it to use significantly less computational power during the fine-tuning and inference processes, which greatly enhances its applicability in real-world scenarios. Notably, Chinchilla achieves a remarkable average accuracy of 67.5% on the MMLU benchmark, marking over a 7% enhancement compared to Gopher, showcasing its superior performance in the field. This impressive capability positions Chinchilla as a leading contender in the realm of language models.

Llama 4 Maverick is a cutting-edge multimodal AI model with 17 billion active parameters and 128 experts, setting a new standard for efficiency and performance. It excels in diverse domains, outperforming other models such as GPT-4o and Gemini 2.0 Flash in coding, reasoning, and image-related tasks. Llama 4 Maverick integrates both text and image processing seamlessly, offering enhanced capabilities for complex tasks such as visual question answering, content generation, and problem-solving. The model’s performance-to-cost ratio makes it an ideal choice for businesses looking to integrate powerful AI into their operations without the hefty resource demands.

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.

Phi-4-mini-reasoning is a transformer-based language model with 3.8 billion parameters, specifically designed to excel in mathematical reasoning and methodical problem-solving within environments that have limited computational capacity or latency constraints. Its optimization stems from fine-tuning with synthetic data produced by the DeepSeek-R1 model, striking a balance between efficiency and sophisticated reasoning capabilities. With training that encompasses over one million varied math problems, ranging in complexity from middle school to Ph.D. level, Phi-4-mini-reasoning demonstrates superior performance to its base model in generating lengthy sentences across multiple assessments and outshines larger counterparts such as OpenThinker-7B, Llama-3.2-3B-instruct, and DeepSeek-R1. Equipped with a 128K-token context window, it also facilitates function calling, which allows for seamless integration with various external tools and APIs. Moreover, Phi-4-mini-reasoning can be quantized through the Microsoft Olive or Apple MLX Framework, enabling its deployment on a variety of edge devices, including IoT gadgets, laptops, and smartphones. Its design not only enhances user accessibility but also expands the potential for innovative applications in mathematical fields.

Mistral 3 represents the newest iteration of open-weight AI models developed by Mistral AI, encompassing a diverse range of models that span from compact, edge-optimized versions to a leading large-scale multimodal model. This lineup features three efficient “Ministral 3” models with 3 billion, 8 billion, and 14 billion parameters, tailored for deployment on devices with limited resources, such as laptops, drones, or other edge devices. Additionally, there is the robust “Mistral Large 3,” which is a sparse mixture-of-experts model boasting a staggering 675 billion total parameters, with 41 billion of them being active. These models are designed to handle multimodal and multilingual tasks, excelling not only in text processing but also in image comprehension, and they have showcased exceptional performance on general queries, multilingual dialogues, and multimodal inputs. Furthermore, both the base and instruction-fine-tuned versions are made available under the Apache 2.0 license, allowing for extensive customization and integration into various enterprise and open-source initiatives. This flexibility in licensing encourages innovation and collaboration among developers and organizations alike.

The Mixtral 8x22B represents our newest open model, establishing a new benchmark for both performance and efficiency in the AI sector. This sparse Mixture-of-Experts (SMoE) model activates only 39B parameters from a total of 141B, ensuring exceptional cost efficiency relative to its scale. Additionally, it demonstrates fluency in multiple languages, including English, French, Italian, German, and Spanish, while also possessing robust skills in mathematics and coding. With its native function calling capability, combined with the constrained output mode utilized on la Plateforme, it facilitates the development of applications and the modernization of technology stacks on a large scale. The model's context window can handle up to 64K tokens, enabling accurate information retrieval from extensive documents. We prioritize creating models that maximize cost efficiency for their sizes, thereby offering superior performance-to-cost ratios compared to others in the community. The Mixtral 8x22B serves as a seamless extension of our open model lineage, and its sparse activation patterns contribute to its speed, making it quicker than any comparable dense 70B model on the market. Furthermore, its innovative design positions it as a leading choice for developers seeking high-performance solutions.

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 are excited to announce the launch of Phi-2, a language model featuring 2.7 billion parameters that excels in reasoning and language comprehension, achieving top-tier results compared to other base models with fewer than 13 billion parameters. In challenging benchmarks, Phi-2 competes with and often surpasses models that are up to 25 times its size, a feat made possible by advancements in model scaling and meticulous curation of training data. Due to its efficient design, Phi-2 serves as an excellent resource for researchers interested in areas such as mechanistic interpretability, enhancing safety measures, or conducting fine-tuning experiments across a broad spectrum of tasks. To promote further exploration and innovation in language modeling, Phi-2 has been integrated into the Azure AI Studio model catalog, encouraging collaboration and development within the research community. Researchers can leverage this model to unlock new insights and push the boundaries of language technology.

Yi-Lightning, a product of 01.AI and spearheaded by Kai-Fu Lee, marks a significant leap forward in the realm of large language models, emphasizing both performance excellence and cost-effectiveness. With the ability to process a context length of up to 16K tokens, it offers an attractive pricing model of $0.14 per million tokens for both inputs and outputs, making it highly competitive in the market. The model employs an improved Mixture-of-Experts (MoE) framework, featuring detailed expert segmentation and sophisticated routing techniques that enhance its training and inference efficiency. Yi-Lightning has distinguished itself across multiple fields, achieving top distinctions in areas such as Chinese language processing, mathematics, coding tasks, and challenging prompts on chatbot platforms, where it ranked 6th overall and 9th in style control. Its creation involved an extensive combination of pre-training, targeted fine-tuning, and reinforcement learning derived from human feedback, which not only enhances its performance but also prioritizes user safety. Furthermore, the model's design includes significant advancements in optimizing both memory consumption and inference speed, positioning it as a formidable contender in its field.

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.