Top Yi-Lightning Alternatives

We are excited to introduce DBRX, a highly adaptable open LLM created by Databricks. This cutting-edge model sets a new standard for open LLMs by achieving remarkable performance across a wide range of established benchmarks. It offers both open-source developers and businesses the advanced features that were traditionally limited to proprietary model APIs; our assessments show that it surpasses GPT-3.5 and stands strong against Gemini 1.0 Pro. Furthermore, DBRX shines as a coding model, outperforming dedicated systems like CodeLLaMA-70B in various programming tasks, while also proving its capability as a general-purpose LLM. The exceptional quality of DBRX is further enhanced by notable improvements in training and inference efficiency. With its sophisticated fine-grained mixture-of-experts (MoE) architecture, DBRX pushes the efficiency of open models to unprecedented levels. In terms of inference speed, it can achieve performance that is twice as fast as LLaMA2-70B, and its total and active parameter counts are around 40% of those found in Grok-1, illustrating its compact structure without sacrificing performance. This unique blend of velocity and size positions DBRX as a transformative force in the realm of open AI models, promising to reshape expectations in the industry. As it continues to evolve, the potential applications for DBRX in various sectors are vast and exciting.

Qwen2.5-Max is a cutting-edge Mixture-of-Experts (MoE) model developed by the Qwen team, trained on a vast dataset of over 20 trillion tokens and improved through techniques such as Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF). It outperforms models like DeepSeek V3 in various evaluations, excelling in benchmarks such as Arena-Hard, LiveBench, LiveCodeBench, and GPQA-Diamond, and also achieving impressive results in tests like MMLU-Pro. Users can access this model via an API on Alibaba Cloud, which facilitates easy integration into various applications, and they can also engage with it directly on Qwen Chat for a more interactive experience. Furthermore, Qwen2.5-Max's advanced features and high performance mark a remarkable step forward in the evolution of AI technology. It not only enhances productivity but also opens new avenues for innovation in the field.

Yi-Large is a cutting-edge proprietary large language model developed by 01.AI, boasting an impressive context length of 32,000 tokens and a pricing model set at $2 per million tokens for both input and output. Celebrated for its exceptional capabilities in natural language processing, common-sense reasoning, and multilingual support, it stands out in competition with leading models like GPT-4 and Claude3 in diverse assessments. The model excels in complex tasks that demand deep inference, precise prediction, and thorough language understanding, making it particularly suitable for applications such as knowledge retrieval, data classification, and the creation of conversational chatbots that closely resemble human communication. Utilizing a decoder-only transformer architecture, Yi-Large integrates advanced features such as pre-normalization and Group Query Attention, having been trained on a vast, high-quality multilingual dataset to optimize its effectiveness. Its versatility and cost-effective pricing make it a powerful contender in the realm of artificial intelligence, particularly for organizations aiming to adopt AI technologies on a worldwide scale. Furthermore, its adaptability across various applications highlights its potential to transform how businesses utilize language models for an array of requirements, paving the way for innovative solutions in the industry. Thus, Yi-Large not only meets but also exceeds expectations, solidifying its role as a pivotal tool in the advancements of AI-driven communication.

DeepSeek-V2 represents an advanced Mixture-of-Experts (MoE) language model created by DeepSeek-AI, recognized for its economical training and superior inference efficiency. This model features a staggering 236 billion parameters, engaging only 21 billion for each token, and can manage a context length stretching up to 128K tokens. It employs sophisticated architectures like Multi-head Latent Attention (MLA) to enhance inference by reducing the Key-Value (KV) cache and utilizes DeepSeekMoE for cost-effective training through sparse computations. When compared to its earlier version, DeepSeek 67B, this model exhibits substantial advancements, boasting a 42.5% decrease in training costs, a 93.3% reduction in KV cache size, and a remarkable 5.76-fold increase in generation speed. With training based on an extensive dataset of 8.1 trillion tokens, DeepSeek-V2 showcases outstanding proficiency in language understanding, programming, and reasoning tasks, thereby establishing itself as a premier open-source model in the current landscape. Its groundbreaking methodology not only enhances performance but also sets unprecedented standards in the realm of artificial intelligence, inspiring future innovations in the field.

ChatGLM-6B is a dialogue model that operates in both Chinese and English, constructed on the General Language Model (GLM) architecture, featuring a robust 6.2 billion parameters. Utilizing advanced model quantization methods, it can efficiently function on typical consumer graphics cards, needing just 6GB of video memory at the INT4 quantization tier. This model incorporates techniques similar to those utilized in ChatGPT but is specifically optimized to improve interactions and dialogues in Chinese. After undergoing rigorous training with around 1 trillion identifiers across both languages, it has also benefited from enhanced supervision, fine-tuning, self-guided feedback, and reinforcement learning driven by human input. As a result, ChatGLM-6B has shown remarkable proficiency in generating responses that resonate effectively with users. Its versatility and high performance render it an essential asset for facilitating bilingual communication, making it an invaluable resource in multilingual environments.

DeepSeek R2 is the much-anticipated successor to the original DeepSeek R1, an AI reasoning model that garnered significant attention upon its launch in January 2025 by the Chinese startup DeepSeek. This latest iteration enhances the impressive groundwork laid by R1, which transformed the AI domain by delivering cost-effective capabilities that rival top-tier models such as OpenAI's o1. R2 is poised to deliver a notable enhancement in performance, promising rapid processing and reasoning skills that closely mimic human capabilities, especially in demanding fields like intricate coding and higher-level mathematics. By leveraging DeepSeek's advanced Mixture-of-Experts framework alongside refined training methodologies, R2 aims to exceed the benchmarks set by its predecessor while maintaining a low computational footprint. Furthermore, there is a strong expectation that this model will expand its reasoning prowess to include additional languages beyond English, potentially enhancing its applicability on a global scale. The excitement surrounding R2 underscores the continuous advancement of AI technology and its potential to impact a variety of sectors significantly, paving the way for innovations that could redefine how we interact with machines.

We are excited to unveil the latest version of our open-source large language model, which includes model weights and initial code for the pretrained and fine-tuned Llama language models, ranging from 7 billion to 70 billion parameters. The Llama 2 pretrained models have been crafted using a remarkable 2 trillion tokens and boast double the context length compared to the first iteration, Llama 1. Additionally, the fine-tuned models have been refined through the insights gained from over 1 million human annotations. Llama 2 showcases outstanding performance compared to various other open-source language models across a wide array of external benchmarks, particularly excelling in reasoning, coding abilities, proficiency, and knowledge assessments. For its training, Llama 2 leveraged publicly available online data sources, while the fine-tuned variant, Llama-2-chat, integrates publicly accessible instruction datasets alongside the extensive human annotations mentioned earlier. Our project is backed by a robust coalition of global stakeholders who are passionate about our open approach to AI, including companies that have offered valuable early feedback and are eager to collaborate with us on Llama 2. The enthusiasm surrounding Llama 2 not only highlights its advancements but also marks a significant transformation in the collaborative development and application of AI technologies. This collective effort underscores the potential for innovation that can emerge when the community comes together to share resources and insights.

Baichuan-13B is a powerful language model featuring 13 billion parameters, created by Baichuan Intelligent as both an open-source and commercially accessible option, and it builds on the previous Baichuan-7B model. This new iteration has excelled in key benchmarks for both Chinese and English, surpassing other similarly sized models in performance. It offers two different pre-training configurations: Baichuan-13B-Base and Baichuan-13B-Chat. Significantly, Baichuan-13B increases its parameter count to 13 billion, utilizing the groundwork established by Baichuan-7B, and has been trained on an impressive 1.4 trillion tokens sourced from high-quality datasets, achieving a 40% increase in training data compared to LLaMA-13B. It stands out as the most comprehensively trained open-source model within the 13B parameter range. Furthermore, it is designed to be bilingual, supporting both Chinese and English, employs ALiBi positional encoding, and features a context window size of 4096 tokens, which provides it with the flexibility needed for a wide range of natural language processing tasks. This model's advancements mark a significant step forward in the capabilities of large language models.

Teuken-7B is a cutting-edge multilingual language model designed to address the diverse linguistic landscape of Europe, emerging from the OpenGPT-X initiative. This model has been trained on a dataset where more than half comprises non-English content, effectively encompassing all 24 official languages of the European Union to ensure robust performance across these tongues. One of the standout features of Teuken-7B is its specially crafted multilingual tokenizer, which has been optimized for European languages, resulting in improved training efficiency and reduced inference costs compared to standard monolingual tokenizers. Users can choose between two distinct versions of the model: Teuken-7B-Base, which offers a foundational pre-trained experience, and Teuken-7B-Instruct, fine-tuned to enhance its responsiveness to user inquiries. Both variations are easily accessible on Hugging Face, promoting transparency and collaboration in the artificial intelligence sector while stimulating further advancements. The development of Teuken-7B not only showcases a commitment to fostering AI solutions but also underlines the importance of inclusivity and representation of Europe's rich cultural tapestry in technology. This initiative ultimately aims to bridge communication gaps and facilitate understanding among diverse populations across the continent.

We are excited to unveil Mistral NeMo, our latest and most sophisticated small model, boasting an impressive 12 billion parameters and a vast context length of 128,000 tokens, all available under the Apache 2.0 license. In collaboration with NVIDIA, Mistral NeMo stands out in its category for its exceptional reasoning capabilities, extensive world knowledge, and coding skills. Its architecture adheres to established industry standards, ensuring it is user-friendly and serves as a smooth transition for those currently using Mistral 7B. To encourage adoption by researchers and businesses alike, we are providing both pre-trained base models and instruction-tuned checkpoints, all under the Apache license. A remarkable feature of Mistral NeMo is its quantization awareness, which enables FP8 inference while maintaining high performance levels. Additionally, the model is well-suited for a range of global applications, showcasing its ability in function calling and offering a significant context window. When benchmarked against Mistral 7B, Mistral NeMo demonstrates a marked improvement in comprehending and executing intricate instructions, highlighting its advanced reasoning abilities and capacity to handle complex multi-turn dialogues. Furthermore, its design not only enhances its performance but also positions it as a formidable option for multi-lingual tasks, ensuring it meets the diverse needs of various use cases while paving the way for future innovations.

Reka Flash 3 stands as a state-of-the-art multimodal AI model, boasting 21 billion parameters and developed by Reka AI, to excel in diverse tasks such as engaging in general conversations, coding, adhering to instructions, and executing various functions. This innovative model skillfully processes and interprets a wide range of inputs, which includes text, images, video, and audio, making it a compact yet versatile solution fit for numerous applications. Constructed from the ground up, Reka Flash 3 was trained on a diverse collection of datasets that include both publicly accessible and synthetic data, undergoing a thorough instruction tuning process with carefully selected high-quality information to refine its performance. The concluding stage of its training leveraged reinforcement learning techniques, specifically the REINFORCE Leave One-Out (RLOO) method, which integrated both model-driven and rule-oriented rewards to enhance its reasoning capabilities significantly. With a remarkable context length of 32,000 tokens, Reka Flash 3 effectively competes against proprietary models such as OpenAI's o1-mini, making it highly suitable for applications that demand low latency or on-device processing. Operating at full precision, the model requires a memory footprint of 39GB (fp16), but this can be optimized down to just 11GB through 4-bit quantization, showcasing its flexibility across various deployment environments. Furthermore, Reka Flash 3's advanced features ensure that it can adapt to a wide array of user requirements, thereby reinforcing its position as a leader in the realm of multimodal AI technology. This advancement not only highlights the progress made in AI but also opens doors to new possibilities for innovation across different sectors.

Tülu 3 represents a state-of-the-art language model designed by the Allen Institute for AI (Ai2) with the objective of enhancing expertise in various domains such as knowledge, reasoning, mathematics, coding, and safety. Built on the foundation of the Llama 3 Base, it undergoes an intricate four-phase post-training process: meticulous prompt curation and synthesis, supervised fine-tuning across a diverse range of prompts and outputs, preference tuning with both off-policy and on-policy data, and a distinctive reinforcement learning approach that bolsters specific skills through quantifiable rewards. This open-source model is distinguished by its commitment to transparency, providing comprehensive access to its training data, coding resources, and evaluation metrics, thus helping to reduce the performance gap typically seen between open-source and proprietary fine-tuning methodologies. Performance evaluations indicate that Tülu 3 excels beyond similarly sized models, such as Llama 3.1-Instruct and Qwen2.5-Instruct, across multiple benchmarks, emphasizing its superior effectiveness. The ongoing evolution of Tülu 3 not only underscores a dedication to enhancing AI capabilities but also fosters an inclusive and transparent technological landscape. As such, it paves the way for future advancements in artificial intelligence that prioritize collaboration and accessibility for all users.

The Mixtral 8x22B is our latest open model, setting a new standard in performance and efficiency within the realm of AI. By utilizing a sparse Mixture-of-Experts (SMoE) architecture, it activates only 39 billion parameters out of a total of 141 billion, leading to remarkable cost efficiency relative to its size. Moreover, it exhibits proficiency in several languages, such as English, French, Italian, German, and Spanish, alongside strong capabilities in mathematics and programming. Its native function calling feature, paired with the constrained output mode used on la Plateforme, greatly aids in application development and the large-scale modernization of technology infrastructures. The model boasts a context window of up to 64,000 tokens, allowing for precise information extraction from extensive documents. We are committed to designing models that optimize cost efficiency, thus providing exceptional performance-to-cost ratios compared to alternatives available in the market. As a continuation of our open model lineage, the Mixtral 8x22B's sparse activation patterns enhance its speed, making it faster than any similarly sized dense 70 billion model available. Additionally, its pioneering design and performance metrics make it an outstanding option for developers in search of high-performance AI solutions, further solidifying its position as a vital asset in the fast-evolving tech landscape.

Qwen2 is a comprehensive array of advanced language models developed by the Qwen team at Alibaba Cloud. This collection includes various models that range from base to instruction-tuned versions, with parameters from 0.5 billion up to an impressive 72 billion, demonstrating both dense configurations and a Mixture-of-Experts architecture. The Qwen2 lineup is designed to surpass many earlier open-weight models, including its predecessor Qwen1.5, while also competing effectively against proprietary models across several benchmarks in domains such as language understanding, text generation, multilingual capabilities, programming, mathematics, and logical reasoning. Additionally, this cutting-edge series is set to significantly influence the artificial intelligence landscape, providing enhanced functionalities that cater to a wide array of applications. As such, the Qwen2 models not only represent a leap in technological advancement but also pave the way for future innovations in the field.

CodeQwen acts as the programming equivalent of Qwen, a collection of large language models developed by the Qwen team at Alibaba Cloud. This model, which is based on a transformer architecture that operates purely as a decoder, has been rigorously pre-trained on an extensive dataset of code. It is known for its strong capabilities in code generation and has achieved remarkable results on various benchmarking assessments. CodeQwen can understand and generate long contexts of up to 64,000 tokens and supports 92 programming languages, excelling in tasks such as text-to-SQL queries and debugging operations. Interacting with CodeQwen is uncomplicated; users can start a dialogue with just a few lines of code leveraging transformers. The interaction is rooted in creating the tokenizer and model using pre-existing methods, utilizing the generate function to foster communication through the chat template specified by the tokenizer. Adhering to our established guidelines, we adopt the ChatML template specifically designed for chat models. This model efficiently completes code snippets according to the prompts it receives, providing responses that require no additional formatting changes, thereby significantly enhancing the user experience. The smooth integration of these components highlights the adaptability and effectiveness of CodeQwen in addressing a wide range of programming challenges, making it an invaluable tool for developers.

DeepSeek-R1 represents a state-of-the-art open-source reasoning model developed by DeepSeek, designed to rival OpenAI's Model o1. Accessible through web, app, and API platforms, it demonstrates exceptional skills in intricate tasks such as mathematics and programming, achieving notable success on exams like the American Invitational Mathematics Examination (AIME) and MATH. This model employs a mixture of experts (MoE) architecture, featuring an astonishing 671 billion parameters, of which 37 billion are activated for every token, enabling both efficient and accurate reasoning capabilities. As part of DeepSeek's commitment to advancing artificial general intelligence (AGI), this model highlights the significance of open-source innovation in the realm of AI. Additionally, its sophisticated features have the potential to transform our methodologies in tackling complex challenges across a variety of fields, paving the way for novel solutions and advancements. The influence of DeepSeek-R1 may lead to a new era in how we understand and utilize AI for problem-solving.

Our sophisticated multimodal assistant has been thoughtfully designed with an emphasis on privacy, security, and operational efficiency. Yasa is equipped to analyze a range of content types, such as text, images, videos, and tables, with ambitions to broaden its capabilities in the future. It serves as a valuable resource for generating ideas for creative endeavors, addressing basic inquiries, and extracting meaningful insights from your proprietary data. With only a few simple commands, you can create, train, compress, or implement it on your own infrastructure. Our unique algorithms allow for customization of the model to suit your individual data and needs. We employ cutting-edge methods that include retrieval, fine-tuning, self-supervised instruction tuning, and reinforcement learning to enhance our model, ensuring it aligns effectively with your specific operational demands. This approach not only improves user satisfaction but also fosters productivity and innovation in a rapidly evolving landscape. As we continue to refine our technology, we remain committed to providing solutions that empower users to achieve their goals.

Falcon-40B is a decoder-only model boasting 40 billion parameters, created by TII and trained on a massive dataset of 1 trillion tokens from RefinedWeb, along with other carefully chosen datasets. It is shared under the Apache 2.0 license, making it accessible for various uses. Why should you consider utilizing Falcon-40B? This model distinguishes itself as the premier open-source choice currently available, outpacing rivals such as LLaMA, StableLM, RedPajama, and MPT, as highlighted by its position on the OpenLLM Leaderboard. Its architecture is optimized for efficient inference and incorporates advanced features like FlashAttention and multiquery functionality, enhancing its performance. Additionally, the flexible Apache 2.0 license allows for commercial utilization without the burden of royalties or limitations. It's essential to recognize that this model is in its raw, pretrained state and is typically recommended to be fine-tuned to achieve the best results for most applications. For those seeking a version that excels in managing general instructions within a conversational context, Falcon-40B-Instruct might serve as a suitable alternative worth considering. Overall, Falcon-40B represents a formidable tool for developers looking to leverage cutting-edge AI technology in their projects.

StarCoder and StarCoderBase are sophisticated Large Language Models crafted for coding tasks, built from freely available data sourced from GitHub, which includes an extensive array of over 80 programming languages, along with Git commits, GitHub issues, and Jupyter notebooks. Similarly to LLaMA, these models were developed with around 15 billion parameters trained on an astonishing 1 trillion tokens. Additionally, StarCoderBase was specifically optimized with 35 billion Python tokens, culminating in the evolution of what we now recognize as StarCoder. Our assessments revealed that StarCoderBase outperforms other open-source Code LLMs when evaluated against well-known programming benchmarks, matching or even exceeding the performance of proprietary models like OpenAI's code-cushman-001 and the original Codex, which was instrumental in the early development of GitHub Copilot. With a remarkable context length surpassing 8,000 tokens, the StarCoder models can manage more data than any other open LLM available, thus unlocking a plethora of possibilities for innovative applications. This adaptability is further showcased by our ability to engage with the StarCoder models through a series of interactive dialogues, effectively transforming them into versatile technical aides capable of assisting with a wide range of programming challenges. Furthermore, this interactive capability enhances user experience, making it easier for developers to obtain immediate support and insights on complex coding issues.

Mistral AI has introduced two advanced models tailored for on-device computing and edge applications, collectively known as "les Ministraux": Ministral 3B and Ministral 8B. These models are particularly remarkable for their abilities in knowledge retention, commonsense reasoning, function-calling, and overall operational efficiency, all while being under the 10B parameter threshold. With support for an impressive context length of up to 128k, they cater to a wide array of applications, including on-device translation, offline smart assistants, local analytics, and autonomous robotics. A standout feature of the Ministral 8B is its incorporation of an interleaved sliding-window attention mechanism, which significantly boosts both the speed and memory efficiency during inference. Both models excel in acting as intermediaries in intricate multi-step workflows, adeptly managing tasks such as input parsing, task routing, and API interactions according to user intentions while keeping latency and operational costs to a minimum. Benchmark results indicate that les Ministraux consistently outperform comparable models across numerous tasks, further cementing their competitive edge in the market. As of October 16, 2024, these innovative models are accessible to developers and businesses, with the Ministral 8B priced competitively at $0.1 per million tokens used. This pricing model promotes accessibility for users eager to incorporate sophisticated AI functionalities into their projects, potentially revolutionizing how AI is utilized in everyday applications.

Ai2 OLMoE is a completely open-source language model that utilizes a mixture-of-experts approach, designed to operate fully on-device, which allows users to explore its capabilities in a secure and private environment. The primary goal of this application is to aid researchers in enhancing on-device intelligence while enabling developers to rapidly prototype innovative AI applications without relying on cloud services. As a highly efficient version within the Ai2 OLMo model family, OLMoE empowers users to engage with advanced local models in practical situations, explore strategies to improve smaller AI systems, and locally test their models using the provided open-source framework. Furthermore, OLMoE can be smoothly integrated into a variety of iOS applications, prioritizing user privacy and security by functioning entirely on-device. Users can easily share the results of their conversations with friends or colleagues, enjoying the benefits of a completely open-source model and application code. This makes Ai2 OLMoE an outstanding resource for personal experimentation and collaborative research, offering extensive opportunities for innovation and discovery in the field of artificial intelligence. By leveraging OLMoE, users can contribute to a growing ecosystem of on-device AI solutions that respect user privacy while facilitating cutting-edge advancements.

Recent advancements in natural language processing, understanding, and generation have largely stemmed from the evolution of large language models. This study introduces a system that utilizes Ascend 910 AI processors alongside the MindSpore framework to train a language model that surpasses one trillion parameters, achieving a total of 1.085 trillion, designated as PanGu-{\Sigma}. This model builds upon the foundation laid by PanGu-{\alpha} by transforming the traditional dense Transformer architecture into a sparse configuration via a technique called Random Routed Experts (RRE). By leveraging an extensive dataset comprising 329 billion tokens, the model was successfully trained with a method known as Expert Computation and Storage Separation (ECSS), which led to an impressive 6.3-fold increase in training throughput through the application of heterogeneous computing. Experimental results revealed that PanGu-{\Sigma} sets a new standard in zero-shot learning for various downstream tasks in Chinese NLP, highlighting its significant potential for progressing the field. This breakthrough not only represents a considerable enhancement in the capabilities of language models but also underscores the importance of creative training methodologies and structural innovations in shaping future developments. As such, this research paves the way for further exploration into improving language model efficiency and effectiveness.

We are excited to unveil an open-source AI model that offers the ability to be fine-tuned, distilled, and deployed across a wide range of platforms. Our latest instruction-tuned model is available in three different sizes: 8B, 70B, and 405B, allowing you to select an option that best fits your unique needs. The open ecosystem we provide accelerates your development journey with a variety of customized product offerings tailored to meet your specific project requirements. You can choose between real-time inference and batch inference services, depending on what your project requires, giving you added flexibility to optimize performance. Furthermore, downloading model weights can significantly enhance cost efficiency per token while you fine-tune the model for your application. To further improve performance, you can leverage synthetic data and seamlessly deploy your solutions either on-premises or in the cloud. By taking advantage of Llama system components, you can also expand the model's capabilities through the use of zero-shot tools and retrieval-augmented generation (RAG), promoting more agentic behaviors in your applications. Utilizing the extensive 405B high-quality data enables you to fine-tune specialized models that cater specifically to various use cases, ensuring that your applications function at their best. In conclusion, this empowers developers to craft innovative solutions that not only meet efficiency standards but also drive effectiveness in their respective domains, leading to a significant impact on the technology landscape.

The Mixtral 8x7B model represents a cutting-edge sparse mixture of experts (SMoE) architecture that features open weights and is made available under the Apache 2.0 license. This innovative model outperforms Llama 2 70B across a range of benchmarks, while also achieving inference speeds that are sixfold faster. As the premier open-weight model with a versatile licensing structure, Mixtral stands out for its impressive cost-effectiveness and performance metrics. Furthermore, it competes with and frequently exceeds the capabilities of GPT-3.5 in many established benchmarks, underscoring its importance in the AI landscape. Its unique blend of accessibility, rapid processing, and overall effectiveness positions it as an attractive option for developers in search of top-tier AI solutions. Consequently, the Mixtral model not only enhances the current technological landscape but also paves the way for future advancements in AI development.

This repository presents the research preview for LongLLaMA, an innovative large language model capable of handling extensive contexts, reaching up to 256,000 tokens or potentially even more. Built on the OpenLLaMA framework, LongLLaMA has been fine-tuned using the Focused Transformer (FoT) methodology. The foundational code for this model comes from Code Llama. We are excited to introduce a smaller 3B base version of the LongLLaMA model, which is not instruction-tuned, and it will be released under an open license (Apache 2.0). Accompanying this release is inference code that supports longer contexts, available on Hugging Face. The model's weights are designed to effortlessly integrate with existing systems tailored for shorter contexts, particularly those that accommodate up to 2048 tokens. In addition to these features, we provide evaluation results and comparisons to the original OpenLLaMA models, thus offering a thorough insight into LongLLaMA's effectiveness in managing long-context tasks. This advancement marks a significant step forward in the field of language models, enabling more sophisticated applications and research opportunities.

The Qwen2.5-1M language model, developed by the Qwen team, is an open-source innovation designed to handle extraordinarily long context lengths of up to one million tokens. This release features two model variations: Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M, marking a groundbreaking milestone as the first Qwen models optimized for such extensive token context. Moreover, the team has introduced an inference framework utilizing vLLM along with sparse attention mechanisms, which significantly boosts processing speeds for inputs of 1 million tokens, achieving speed enhancements ranging from three to seven times. Accompanying this model is a comprehensive technical report that delves into the design decisions and outcomes of various ablation studies. This thorough documentation ensures that users gain a deep understanding of the models' capabilities and the technology that powers them. Additionally, the improvements in processing efficiency are expected to open new avenues for applications needing extensive context management.

Pre-trained language models have advanced significantly, demonstrating exceptional performance in various Natural Language Processing (NLP) tasks. The remarkable features of GPT-3 illustrate that scaling these models can lead to the discovery of their immense capabilities. Recently, the introduction of a comprehensive framework called ERNIE 3.0 has allowed for the pre-training of large-scale models infused with knowledge, resulting in a model with an impressive 10 billion parameters. This version of ERNIE 3.0 has outperformed many leading models across numerous NLP challenges. In our pursuit of exploring the impact of scaling, we have created an even larger model named ERNIE 3.0 Titan, which boasts up to 260 billion parameters and is developed on the PaddlePaddle framework. Moreover, we have incorporated a self-supervised adversarial loss coupled with a controllable language modeling loss, which empowers ERNIE 3.0 Titan to generate text that is both accurate and adaptable, thus extending the limits of what these models can achieve. This innovative methodology not only improves the model's overall performance but also paves the way for new research opportunities in the fields of text generation and fine-tuning control. As the landscape of NLP continues to evolve, the advancements in these models promise to drive further breakthroughs in understanding and generating human language.

A sophisticated End-to-End MLLM has been developed to accommodate various types of references and effectively ground its responses. The Ferret Model employs a unique combination of Hybrid Region Representation and a Spatial-aware Visual Sampler, which facilitates detailed and adaptable referring and grounding functions within the MLLM framework. Serving as a foundational element, the GRIT Dataset consists of about 1.1 million entries, specifically designed as a large-scale and hierarchical dataset aimed at enhancing instruction tuning in the ground-and-refer domain. Moreover, the Ferret-Bench acts as a thorough multimodal evaluation benchmark that concurrently measures referring, grounding, semantics, knowledge, and reasoning, thus providing a comprehensive assessment of the model's performance. This elaborate configuration is intended to improve the synergy between language and visual information, which could lead to more intuitive AI systems that better understand and interact with users. Ultimately, advancements in these models may significantly transform how we engage with technology in our daily lives.

In tribute to Cleopatra, whose dramatic story ended with the fateful encounter with a snake, we proudly present Codestral Mamba, a Mamba2 language model tailored for code generation and made available under an Apache 2.0 license. Codestral Mamba marks a pivotal step forward in our commitment to pioneering and refining innovative architectures. This model is available for free use, modification, and distribution, and we hope it will pave the way for new discoveries in architectural research. The Mamba models stand out due to their linear time inference capabilities, coupled with a theoretical ability to manage sequences of infinite length. This unique characteristic allows users to engage with the model seamlessly, delivering quick responses irrespective of the input size. Such remarkable efficiency is especially beneficial for boosting coding productivity; hence, we have integrated advanced coding and reasoning abilities into this model, ensuring it can compete with top-tier transformer-based models. As we push the boundaries of innovation, we are confident that Codestral Mamba will not only advance coding practices but also inspire new generations of developers. This exciting release underscores our dedication to fostering creativity and productivity within the tech community.

We provide rapid and accurate AI models tailored for effective use in production settings. Our inference API is engineered for maximum uptime, harnessing the latest NVIDIA GPUs to deliver peak performance. Additionally, we have compiled a diverse array of high-quality open-source natural language processing (NLP) models sourced from the community, making them easily accessible for your projects. You can also customize your own models, including GPT-J, or upload your proprietary models for smooth integration into production. Through a user-friendly dashboard, you can swiftly upload or fine-tune AI models, enabling immediate deployment without the complexities of managing factors like memory constraints, uptime, or scalability. You have the freedom to upload an unlimited number of models and deploy them as necessary, fostering a culture of continuous innovation and adaptability to meet your dynamic needs. This comprehensive approach provides a solid foundation for utilizing AI technologies effectively in your initiatives, promoting growth and efficiency in your workflows.

The Falcon-7B model is a causal decoder-only architecture with a total of 7 billion parameters, created by TII, and trained on a vast dataset consisting of 1,500 billion tokens from RefinedWeb, along with additional carefully curated corpora, all under the Apache 2.0 license. What are the benefits of using Falcon-7B? This model excels compared to other open-source options like MPT-7B, StableLM, and RedPajama, primarily because of its extensive training on an unimaginably large dataset of 1,500 billion tokens from RefinedWeb, supplemented by thoughtfully selected content, which is clearly reflected in its performance ranking on the OpenLLM Leaderboard. Furthermore, it features an architecture optimized for rapid inference, utilizing advanced technologies such as FlashAttention and multiquery strategies. In addition, the flexibility offered by the Apache 2.0 license allows users to pursue commercial ventures without worrying about royalties or stringent constraints. This unique blend of high performance and operational freedom positions Falcon-7B as an excellent option for developers in search of sophisticated modeling capabilities. Ultimately, the model's design and resourcefulness make it a compelling choice in the rapidly evolving landscape of machine learning.

ALBERT is a groundbreaking Transformer model that employs self-supervised learning and has been pretrained on a vast array of English text. Its automated mechanisms remove the necessity for manual data labeling, allowing the model to generate both inputs and labels straight from raw text. The training of ALBERT revolves around two main objectives. The first is Masked Language Modeling (MLM), which randomly masks 15% of the words in a sentence, prompting the model to predict the missing words. This approach stands in contrast to RNNs and autoregressive models like GPT, as it allows for the capture of bidirectional representations in sentences. The second objective, Sentence Ordering Prediction (SOP), aims to ascertain the proper order of two adjacent segments of text during the pretraining process. By implementing these strategies, ALBERT significantly improves its comprehension of linguistic context and structure. This innovative architecture positions ALBERT as a strong contender in the realm of natural language processing, pushing the boundaries of what language models can achieve.

Atla's Selene 1 API introduces state-of-the-art AI evaluation models, enabling developers to establish individualized assessment criteria for accurately measuring the effectiveness of their AI applications. This advanced model outperforms top competitors on well-regarded evaluation benchmarks, ensuring reliable and precise assessments. Users can customize their evaluation processes to meet specific needs through the Alignment Platform, which facilitates in-depth analysis and personalized scoring systems. Beyond providing actionable insights and accurate evaluation metrics, this API seamlessly integrates into existing workflows, enhancing usability. It incorporates established performance metrics, including relevance, correctness, helpfulness, faithfulness, logical coherence, and conciseness, addressing common evaluation issues such as detecting hallucinations in retrieval-augmented generation contexts or comparing outcomes with verified ground truth data. Additionally, the API's adaptability empowers developers to continually innovate and improve their evaluation techniques, making it an essential asset for boosting the performance of AI applications while fostering a culture of ongoing enhancement.

Large language models, which often demand significant computational power and prolonged training periods, have shown remarkable abilities in performing zero- and few-shot learning tasks. The substantial resources required for their creation make it quite difficult for many researchers to replicate these models. Moreover, access to the limited number of models available through APIs is restricted, as users are unable to acquire the full model weights, which hinders academic research. To address these issues, we present Open Pre-trained Transformers (OPT), a series of decoder-only pre-trained transformers that vary in size from 125 million to 175 billion parameters, which we aim to share fully and responsibly with interested researchers. Our research reveals that OPT-175B achieves performance levels comparable to GPT-3, while consuming only one-seventh of the carbon emissions needed for GPT-3's training process. In addition to this, we plan to offer a comprehensive logbook detailing the infrastructural challenges we faced during the project, along with code to aid experimentation with all released models, ensuring that scholars have the necessary resources to further investigate this technology. This initiative not only democratizes access to advanced models but also encourages sustainable practices in the field of artificial intelligence.

We are thrilled to unveil our new X1 large series of models, marking a significant advancement in our offerings. The most powerful model from Giga ML is now available for both pre-training and fine-tuning in an on-premises setup. Our integration with Open AI ensures seamless compatibility with existing tools such as long chain, llama-index, and more, enhancing usability. Additionally, users have the option to pre-train LLMs using tailored data sources, including industry-specific documents or proprietary company files. As the realm of large language models (LLMs) continues to rapidly advance, it presents remarkable opportunities for breakthroughs in natural language processing across diverse sectors. However, the industry still faces several substantial challenges that need addressing. At Giga ML, we are proud to present the X1 Large 32k model, an innovative on-premise LLM solution crafted to confront these key challenges head-on, empowering organizations to fully leverage the capabilities of LLMs. This launch is not just a step forward for our technology, but a major stride towards enhancing the language processing capabilities of businesses everywhere. We believe that by providing these advanced tools, we can drive meaningful improvements in how organizations communicate and operate.

Stability AI, in collaboration with its CarperAI lab, proudly introduces Stable Beluga 1 and its enhanced version, Stable Beluga 2, formerly called FreeWilly, both of which are powerful new Large Language Models (LLMs) now accessible to the public. These innovations demonstrate exceptional reasoning abilities across a diverse array of benchmarks, highlighting their adaptability and robustness. Stable Beluga 1 is constructed upon the foundational LLaMA 65B model and has been carefully fine-tuned using a cutting-edge synthetically-generated dataset through Supervised Fine-Tune (SFT) in the traditional Alpaca format. Similarly, Stable Beluga 2 is based on the LLaMA 2 70B model, further advancing performance standards in the field. The introduction of these models signifies a major advancement in the progression of open access AI technology, paving the way for future developments in the sector. With their release, users can expect enhanced capabilities that could revolutionize various applications.

VideoPoet is a groundbreaking modeling approach that enables any autoregressive language model or large language model (LLM) to function as a powerful video generator. This technique consists of several simple components. An autoregressive language model is trained to understand various modalities—including video, image, audio, and text—allowing it to predict the next video or audio token in a given sequence. The training structure for the LLM includes diverse multimodal generative learning objectives, which encompass tasks like text-to-video, text-to-image, image-to-video, video frame continuation, inpainting and outpainting of videos, video stylization, and video-to-audio conversion. Moreover, these tasks can be integrated to improve the model's zero-shot capabilities. This clear and effective methodology illustrates that language models can not only generate but also edit videos while maintaining impressive temporal coherence, highlighting their potential for sophisticated multimedia applications. Consequently, VideoPoet paves the way for a plethora of new opportunities in creative expression and automated content development, expanding the boundaries of how we produce and interact with digital media.

Qwen-7B represents the seventh iteration in Alibaba Cloud's Qwen language model lineup, also referred to as Tongyi Qianwen, featuring 7 billion parameters. This advanced language model employs a Transformer architecture and has undergone pretraining on a vast array of data, including web content, literature, programming code, and more. In addition, we have launched Qwen-7B-Chat, an AI assistant that enhances the pretrained Qwen-7B model by integrating sophisticated alignment techniques. The Qwen-7B series includes several remarkable attributes: Its training was conducted on a premium dataset encompassing over 2.2 trillion tokens collected from a custom assembly of high-quality texts and codes across diverse fields, covering both general and specialized areas of knowledge. Moreover, the model excels in performance, outshining similarly-sized competitors on various benchmark datasets that evaluate skills in natural language comprehension, mathematical reasoning, and programming challenges. This establishes Qwen-7B as a prominent contender in the AI language model landscape. In summary, its intricate training regimen and solid architecture contribute significantly to its outstanding adaptability and efficiency in a wide range of applications.

PanGu-α is developed with the MindSpore framework and is powered by an impressive configuration of 2048 Ascend 910 AI processors during its training phase. This training leverages a sophisticated parallelism approach through MindSpore Auto-parallel, utilizing five distinct dimensions of parallelism: data parallelism, operation-level model parallelism, pipeline model parallelism, optimizer model parallelism, and rematerialization, to efficiently allocate tasks among the 2048 processors. To enhance the model's generalization capabilities, we compiled an extensive dataset of 1.1TB of high-quality Chinese language information from various domains for pretraining purposes. We rigorously test PanGu-α's generation capabilities across a variety of scenarios, including text summarization, question answering, and dialogue generation. Moreover, we analyze the impact of different model scales on few-shot performance across a broad spectrum of Chinese NLP tasks. Our experimental findings underscore the remarkable performance of PanGu-α, illustrating its proficiency in managing a wide range of tasks, even in few-shot or zero-shot situations, thereby demonstrating its versatility and durability. This thorough assessment not only highlights the strengths of PanGu-α but also emphasizes its promising applications in practical settings. Ultimately, the results suggest that PanGu-α could significantly advance the field of natural language processing.

The newest version of the open-source AI framework, which can be customized and utilized across different platforms, is available in several configurations: 1B, 3B, 11B, and 90B, while still offering the option to use Llama 3.1. Llama 3.2 includes a selection of large language models (LLMs) that are pretrained and fine-tuned specifically for multilingual text processing in 1B and 3B sizes, whereas the 11B and 90B models support both text and image inputs, generating text outputs. This latest release empowers users to build highly effective applications that cater to specific requirements. For applications running directly on devices, such as summarizing conversations or managing calendars, the 1B or 3B models are excellent selections. On the other hand, the 11B and 90B models are particularly suited for tasks involving images, allowing users to manipulate existing pictures or glean further insights from images in their surroundings. Ultimately, this broad spectrum of models opens the door for developers to experiment with creative applications across a wide array of fields, enhancing the potential for innovation and impact.

Hippocratic AI stands as a groundbreaking innovation in the realm of artificial intelligence, outperforming GPT-4 in 105 out of 114 healthcare-related assessments and certifications. Remarkably, it surpassed GPT-4 by at least five percent on 74 of these certifications, with a margin of ten percent or more in 43 instances. Unlike many language models that draw from a wide array of internet resources—which may sometimes lead to the dissemination of incorrect information—Hippocratic AI is focused on obtaining evidence-based healthcare content through legitimate channels. To enhance the model’s efficacy and ensure safety, we are deploying a tailored Reinforcement Learning with Human Feedback approach that actively engages healthcare professionals in both training and validating the model before it reaches the public. This thorough methodology, referred to as RLHF-HP, ensures that Hippocratic AI will be introduced only after receiving endorsement from a considerable number of licensed healthcare experts, emphasizing patient safety and precision in its functionalities. This commitment to stringent validation not only distinguishes Hippocratic AI in the competitive landscape of AI healthcare solutions but also reinforces the trust that users can place in its capabilities. Ultimately, Hippocratic AI sets a new standard for reliability and effectiveness in the field of healthcare technology.

Mistral Small 3.1 is an advanced, multimodal, and multilingual AI model that has been made available under the Apache 2.0 license. Building upon the previous Mistral Small 3, this updated version showcases improved text processing abilities and enhanced multimodal understanding, with the capacity to handle an extensive context window of up to 128,000 tokens. It outperforms comparable models like Gemma 3 and GPT-4o Mini, reaching remarkable inference rates of 150 tokens per second. Designed for versatility, Mistral Small 3.1 excels in various applications, including instruction adherence, conversational interaction, visual data interpretation, and executing functions, making it suitable for both commercial and individual AI uses. Its efficient architecture allows it to run smoothly on hardware configurations such as a single RTX 4090 or a Mac with 32GB of RAM, enabling on-device operations. Users have the option to download the model from Hugging Face and explore its features via Mistral AI's developer playground, while it is also embedded in services like Google Cloud Vertex AI and accessible on platforms like NVIDIA NIM. This extensive flexibility empowers developers to utilize its advanced capabilities across a wide range of environments and applications, thereby maximizing its potential impact in the AI landscape. Furthermore, Mistral Small 3.1's innovative design ensures that it remains adaptable to future technological advancements.

Mistral AI has introduced two state-of-the-art models aimed at on-device computing and edge applications, collectively known as "les Ministraux": Ministral 3B and Ministral 8B. These advanced models set new benchmarks for knowledge, commonsense reasoning, function-calling, and efficiency in the sub-10B category. They offer remarkable flexibility for a variety of applications, from overseeing complex workflows to creating specialized task-oriented agents. With the capability to manage an impressive context length of up to 128k (currently supporting 32k on vLLM), Ministral 8B features a distinctive interleaved sliding-window attention mechanism that boosts both speed and memory efficiency during inference. Crafted for low-latency and compute-efficient applications, these models thrive in environments such as offline translation, internet-independent smart assistants, local data processing, and autonomous robotics. Additionally, when integrated with larger language models like Mistral Large, les Ministraux can serve as effective intermediaries, enhancing function-calling within detailed multi-step workflows. This synergy not only amplifies performance but also extends the potential of AI in edge computing, paving the way for innovative solutions in various fields. The introduction of these models marks a significant step forward in making advanced AI more accessible and efficient for real-world applications.

ChatGPT, developed by OpenAI, is a sophisticated language model that generates coherent and contextually appropriate replies by drawing from a wide selection of internet text. Its extensive training equips it to tackle a multitude of tasks in natural language processing, such as engaging in dialogues, responding to inquiries, and producing text in diverse formats. Leveraging deep learning algorithms, ChatGPT employs a transformer architecture that has demonstrated remarkable efficiency in numerous NLP tasks. Additionally, the model can be customized for specific applications, such as language translation, text categorization, and answering questions, allowing developers to create advanced NLP systems with greater accuracy. Besides its text generation capabilities, ChatGPT is also capable of interpreting and writing code, highlighting its adaptability in managing various content types. This broad range of functionalities not only enhances its utility but also paves the way for innovative integrations into an array of technological solutions. The ongoing advancements in AI technology are likely to further elevate the capabilities of models like ChatGPT, making them even more integral to our everyday interactions with machines.

We are thrilled to introduce MPT-7B, the latest model in the MosaicML Foundation Series. This transformer model has been carefully developed from scratch, utilizing 1 trillion tokens of varied text and code during its training. It is accessible as open-source software, making it suitable for commercial use and achieving performance levels comparable to LLaMA-7B. The entire training process was completed in just 9.5 days on the MosaicML platform, with no human intervention, and incurred an estimated cost of $200,000. With MPT-7B, users can train, customize, and deploy their own versions of MPT models, whether they opt to start from one of our existing checkpoints or initiate a new project. Additionally, we are excited to unveil three specialized variants alongside the core MPT-7B: MPT-7B-Instruct, MPT-7B-Chat, and MPT-7B-StoryWriter-65k+, with the latter featuring an exceptional context length of 65,000 tokens for generating extensive content. These new offerings greatly expand the horizons for developers and researchers eager to harness the capabilities of transformer models in their innovative initiatives. Furthermore, the flexibility and scalability of MPT-7B are designed to cater to a wide range of application needs, fostering creativity and efficiency in developing advanced AI solutions.

Grok 3 Think, the latest iteration of xAI's AI model, seeks to enhance reasoning capabilities by employing advanced reinforcement learning methods. It can tackle complex problems for time spans that range from a few seconds to several minutes, improving its outputs by reviewing earlier steps, exploring alternative solutions, and refining its methods. The model is built on an extraordinary scale, demonstrating remarkable skill across a variety of tasks such as mathematics, programming, and general knowledge, and it has achieved significant results in competitions like the American Invitational Mathematics Examination. Furthermore, Grok 3 Think not only provides accurate responses but also prioritizes transparency, allowing users to explore the reasoning behind its answers, which sets a new standard for artificial intelligence in tackling intricate challenges. By focusing on transparency and reasoning, this model enhances user confidence in AI systems and fosters a deeper understanding of the decision-making mechanics involved. As a result, Grok 3 Think not only excels in performance but also cultivates a more informed user experience regarding AI capabilities.

CodeGemma is an impressive collection of efficient and adaptable models that can handle a variety of coding tasks, such as middle code completion, code generation, natural language processing, mathematical reasoning, and instruction following. It includes three unique model variants: a 7B pre-trained model intended for code completion and generation using existing code snippets, a fine-tuned 7B version for converting natural language queries into code while following instructions, and a high-performing 2B pre-trained model that completes code at speeds up to twice as fast as its counterparts. Whether you are filling in lines, creating functions, or assembling complete code segments, CodeGemma is designed to assist you in any environment, whether local or utilizing Google Cloud services. With its training grounded in a vast dataset of 500 billion tokens, primarily in English and taken from web sources, mathematics, and programming languages, CodeGemma not only improves the syntactical precision of the code it generates but also guarantees its semantic accuracy, resulting in fewer errors and a more efficient debugging process. Beyond just functionality, this powerful tool consistently adapts and improves, making coding more accessible and streamlined for developers across the globe, thereby fostering a more innovative programming landscape. As the technology advances, users can expect even more enhancements in terms of speed and accuracy.

Sky-T1-32B-Preview represents a groundbreaking open-source reasoning model developed by the NovaSky team at UC Berkeley's Sky Computing Lab. It achieves performance levels similar to those of proprietary models like o1-preview across a range of reasoning and coding tests, all while being created for under $450, emphasizing its potential to provide advanced reasoning skills at a lower cost. Fine-tuned from Qwen2.5-32B-Instruct, this model was trained on a carefully selected dataset of 17,000 examples that cover diverse areas, including mathematics and programming. The training was efficiently completed in a mere 19 hours with the aid of eight H100 GPUs using DeepSpeed Zero-3 offloading technology. Notably, every aspect of this project—spanning data, code, and model weights—is fully open-source, enabling both the academic and open-source communities to not only replicate but also enhance the model's functionalities. Such openness promotes a spirit of collaboration and innovation within the artificial intelligence research and development landscape, inviting contributions from various sectors. Ultimately, this initiative represents a significant step forward in making powerful AI tools more accessible to a wider audience.

The multilingual T5 (mT5) is an exceptionally adaptable pretrained text-to-text transformer model, created using a methodology similar to that of the original T5. This repository provides essential resources for reproducing the results detailed in the mT5 research publication. mT5 has undergone training on the vast mC4 corpus, which includes a remarkable 101 languages, such as Afrikaans, Albanian, Amharic, Arabic, Armenian, Azerbaijani, Basque, Belarusian, Bengali, Bulgarian, Burmese, Catalan, Cebuano, Chichewa, Chinese, Corsican, Czech, Danish, Dutch, English, Esperanto, Estonian, Filipino, Finnish, French, Galician, Georgian, German, Greek, Gujarati, Haitian Creole, Hausa, Hawaiian, Hebrew, Hindi, Hmong, Hungarian, Icelandic, Igbo, Indonesian, Irish, Italian, Japanese, Javanese, Kannada, Kazakh, Khmer, Korean, Kurdish, Kyrgyz, Lao, Latin, Latvian, Lithuanian, Luxembourgish, Macedonian, Malagasy, Malay, Malayalam, Maltese, Maori, Marathi, Mongolian, Nepali, Norwegian, Pashto, Persian, Polish, Portuguese, Punjabi, Romanian, Russian, Samoan, Scottish Gaelic, Serbian, Shona, Sindhi, and many more. This extensive language coverage renders mT5 an invaluable asset for multilingual applications in diverse sectors, enhancing its usefulness for researchers and developers alike.

OLMo 2 is a suite of fully open language models developed by the Allen Institute for AI (AI2), designed to provide researchers and developers with straightforward access to training datasets, open-source code, reproducible training methods, and extensive evaluations. These models are trained on a remarkable dataset consisting of up to 5 trillion tokens and are competitive with leading open-weight models such as Llama 3.1, especially in English academic assessments. A significant emphasis of OLMo 2 lies in maintaining training stability, utilizing techniques to reduce loss spikes during prolonged training sessions, and implementing staged training interventions to address capability weaknesses in the later phases of pretraining. Furthermore, the models incorporate advanced post-training methodologies inspired by AI2's Tülu 3, resulting in the creation of OLMo 2-Instruct models. To support continuous enhancements during the development lifecycle, an actionable evaluation framework called the Open Language Modeling Evaluation System (OLMES) has been established, featuring 20 benchmarks that assess vital capabilities. This thorough methodology not only promotes transparency but also actively encourages improvements in the performance of language models, ensuring they remain at the forefront of AI advancements. Ultimately, OLMo 2 aims to empower the research community by providing resources that foster innovation and collaboration in language modeling.