DeepSeek Unveils R1-Lite-Preview: A New Benchmark in AI Reasoning

Deepseek-R1-Lite Open Source LLM Fully Tested

Have you ever found yourself wrestling with a coding problem that just wouldn't budge or staring at a complex equation, wishing for a bit of extra brainpower? If so, you're not alone. Whether you're a developer, researcher, or someone who simply loves solving intricate puzzles, the struggle to find tools that are both powerful and accessible is all too familiar. That's where Deepseek-R1-Lite steps in. This open-source reasoning model isn't just another AI tool -- it's a fantastic option designed to tackle challenges that demand deep thought, precision, and adaptability. From generating clean, functional code to cracking advanced mathematical problems, it promises to make your workflow smoother and your results sharper. What sets Deepseek-R1-Lite apart isn't just its impressive capabilities but its commitment to accessibility. Unlike many proprietary models that lock innovation behind paywalls, this model offers open weights and APIs, empowering anyone to tap into its potential. Imagine having a tool that not only outperforms industry giants like OpenAI's o1 Preview and Claude 3.5 in key benchmarks but also invites you to explore, experiment, and create without barriers. Curious about how it achieves all this? Let's dive into the details and uncover what makes Deepseek-R1-Lite a standout in the world of open-source AI. Deepseek-R1-Lite is purpose-built for reasoning-intensive tasks, excelling in areas that demand logical deduction, long-context generation, and dynamic input handling. Its design emphasizes precision, adaptability, and versatility, making it a valuable tool for a wide range of applications, from algorithm development to creative problem-solving. These features position Deepseek-R1-Lite as a versatile and reliable tool for professionals across various domains. Preliminary testing suggests that Deepseek-R1-Lite outperforms leading proprietary models, including OpenAI's 01 Preview and Claude 3.5, in several critical areas. Its strengths lie in logical reasoning, algorithm design, and error handling, where it has demonstrated superior performance. For example, the model successfully implemented complex algorithms like Bellman-Ford and generated accurate pseudo-code for computational tasks. While official benchmarks are yet to be released, early results indicate that Deepseek-R1-Lite has the potential to set new standards in open-source AI performance. Its ability to handle intricate tasks with precision and consistency makes it a promising alternative to proprietary solutions. Deepseek-R1-Lite's versatility makes it a valuable asset for developers, researchers, educators, and other professionals. Its wide-ranging capabilities enable it to address diverse challenges with efficiency and accuracy. These applications highlight the model's adaptability and its ability to meet the unique needs of various industries. Deepseek-R1-Lite's primary strengths lie in its open-source accessibility, advanced reasoning capabilities, and adaptability to real-world scenarios. Its ability to handle dynamic inputs and error-prone coding tasks further enhances its utility. By offering free access to its open weights and APIs, the model provide widespread access tos AI, making powerful tools available to a broader audience. However, some areas require refinement. For instance, while the model excels in front-end development and algorithm design, its outputs in backend development contexts may need additional fine-tuning. Additionally, the absence of official benchmarks leaves room for further validation, though early results are promising. To access Deepseek-R1-Lite, visit Deepseek's official website. The model features an intuitive "Deep Think" mode, optimized for reasoning-intensive tasks. Developers can integrate its capabilities into their projects using the provided APIs, fostering innovation and collaboration within the open-source community. Deepseek-R1-Lite represents a significant advancement in open-source AI, challenging the dominance of proprietary models by offering powerful reasoning and coding capabilities at no cost. Its open-access approach not only provide widespread access tos AI but also encourages cross-industry innovation and collaboration. Whether you're a developer seeking to optimize workflows, a researcher tackling advanced computational problems, or an educator explaining abstract concepts, Deepseek-R1-Lite provides the tools and flexibility to meet your needs. By bridging the gap between accessibility and performance, it sets a new precedent for what open-source AI can achieve.

DeepSeek's first reasoning model R1-Lite-Preview turns heads, beating OpenAI o1 performance

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More DeepSeek, an open source focused AI offshoot of Chinese quantitative hedge fund High-Flyer Capital Management, has unveiled the R1-Lite-Preview, its latest reasoning-focused large language model, available for now exclusively through DeepSeek Chat, its web-based AI chatbot. Known for its innovative contributions to the open-source AI ecosystem, DeepSeek's new release aims to bring high-level reasoning capabilities to the public while maintaining its commitment to accessible and transparent AI. And the R1-Lite-Preview, despite only being available through the chat application for now, is already turning heads by offering performance nearing and in some cases exceeding OpenAI's vaunted o1-preview model. Like that model released in September 2024, DeepSeek-R1-Lite-Preview exhibits "chain-of-thought" reasoning, showing the user the different chains or trains of "thought" it goes down to respond to their queries and inputs, documenting the process by explaining what it is doing and why. While some of the chains/trains of thoughts may appear nonsensical or even erroneous to humans, DeepSeek-R1-Lite-Preview appears on the whole to be strikingly accurate, even answering "trick" questions that have tripped up other, older, yet powerful AI models such as GPT-4o and Claude's Anthropic family, including "how many letter Rs are in the word Strawberry?" and "which is larger, 9.11 or 9.9?" See screenshots below of my tests of these prompts on DeepSeek Chat: A New Approach to AI Reasoning DeepSeek-R1-Lite-Preview is designed to excel in tasks requiring logical inference, mathematical reasoning, and real-time problem-solving. According to DeepSeek, the model exceeds OpenAI o1-preview-level performance on established benchmarks such as AIME (American Invitational Mathematics Examination) and MATH. Its reasoning capabilities are enhanced by its transparent thought process, allowing users to follow along as the model tackles complex challenges step by step. DeepSeek has also published scaling data, showcasing steady accuracy improvements when the model is given more time or "thought tokens" to solve problems. Performance graphs highlight its proficiency in achieving higher scores on benchmarks such as AIME as thought depth increases. Benchmarks and Real-World Applications DeepSeek-R1-Lite-Preview has performed competitively on key benchmarks. The company's published results highlight its ability to handle a wide range of tasks, from complex mathematics to logic-based scenarios, earning performance scores that rival top-tier models in reasoning benchmarks like GPQA and Codeforces. The transparency of its reasoning process further sets it apart. Users can observe the model's logical steps in real-time, adding an element of accountability and trust that many proprietary AI systems lack. However, DeepSeek has not yet released the full code for independent third-party analysis or benchmarking, nor has it yet made DeepSeek-R1-Lite-Preview available through an API which would allow the same kind of independent tests. In addition, the company has not yet published a blog post nor a technical paper explaining how DeepSeek-R1-Lite-Preview was trained or architected, leaving many question marks about its underlying origins. Accessibility and Open-Source Plans The R1-Lite-Preview is now accessible through DeepSeek Chat at chat.deepseek.com. While free for public use, the model's advanced "Deep Think" mode has a daily limit of 50 messages, offering ample opportunity for users to experience its capabilities. Looking ahead, DeepSeek plans to release open-source versions of its R1 series models and related APIs, according to the company's posts on X. This move aligns with the company's history of supporting the open-source AI community. Its previous release, DeepSeek-V2.5, earned praise for combining general language processing and advanced coding capabilities, making it one of the most powerful open-source AI models at the time. Building on a Legacy DeepSeek is continuing its tradition of pushing boundaries in open-source AI. Earlier models like DeepSeek-V2.5 and DeepSeek Coder demonstrated impressive capabilities across language and coding tasks, with benchmarks placing it as a leader in the field. The release of R1-Lite-Preview adds a new dimension, focusing on transparent reasoning and scalability. As businesses and researchers explore applications for reasoning-intensive AI, DeepSeek's commitment to openness ensures that its models remain a vital resource for development and innovation. By combining high performance, transparent operations, and open-source accessibility, DeepSeek is not just advancing AI but also reshaping how it is shared and used. The R1-Lite-Preview is available now for public testing. Open-source models and APIs are expected to follow, further solidifying DeepSeek's position as a leader in accessible, advanced AI technologies.

How DeepSeek AI is Outperforming Industry Giants Like OpenAI

In a significant development for the artificial intelligence sector, DeepSeek AI, an emerging Chinese tech company, has unveiled its latest model, DeepSeek-R1-Lite. This powerful AI system has reportedly outperformed OpenAI's o1 preview across several critical benchmarks, signaling a notable shift in the AI landscape. Founded just last year in 2023, DeepSeek AI's rapid ascent underscores the dynamic and competitive nature of AI development. The company's latest model excels in complex tasks like coding, mathematics, and natural language processing, showcasing an impressive blend of innovation and efficiency. With advanced techniques like test time compute and majority voting, DeepSeek-R1-Lite is setting new standards for performance and reliability. So, what does this mean for the future of AI? Let's explore how DeepSeek AI's achievements might just be the fantastic option for a new era of innovation and competition in the industry. DeepSeek AI has swiftly established itself as a formidable player in AI model development, tackling a wide array of complex tasks including: The DeepSeek-R1-Lite model has demonstrated exceptional performance in rigorous benchmarks such as AIM and Math 500, surpassing the capabilities of OpenAI's o1 preview. This success can be attributed to the model's innovative architecture and its strategic use of test time compute and majority voting mechanisms, which significantly enhance its computational efficiency and decision-making accuracy. At the heart of DeepSeek-R1-Lite's impressive capabilities lies its utilization of innovative AI techniques: Test Time Compute: This method optimizes computational resources during the testing phase, allowing faster and more accurate results. By efficiently allocating processing power where it's needed most, DeepSeek-R1-Lite can tackle complex problems with remarkable speed and precision. Majority Voting: This approach aggregates multiple outputs to form a final decision, significantly enhancing the model's reliability and accuracy. By considering various potential solutions and selecting the most consistent one, DeepSeek-R1-Lite achieves a higher level of precision in its responses. The synergy between these techniques is crucial to DeepSeek-R1-Lite's ability to outperform competitors in demanding benchmarks, showcasing its potential to handle real-world applications with unprecedented efficiency. Uncover more insights about AI model development in previous articles we have written. A key factor in DeepSeek-R1-Lite's success is its exceptional scalability and problem-solving capabilities. The model's architecture is designed to support efficient scaling, allowing it to tackle increasingly complex tasks without compromising performance. This scalability is particularly evident in its ability to handle diverse challenges, from intricate mathematical computations to nuanced language understanding. The model's advanced problem-solving skills are clearly demonstrated in benchmark results, where it consistently outperforms the o1 preview in tasks requiring mathematical prowess and logical reasoning. This capability positions DeepSeek-R1-Lite as a versatile tool for a wide range of applications, from scientific research to business analytics. One of DeepSeek-R1-Lite's most intriguing features is its implementation of "chains of thought." This innovative approach provides users with insights into the model's reasoning process, offering a window into how the AI arrives at its conclusions. By examining these thought chains, users can: This transparency not only enhances the model's usability but also contributes to the broader field of explainable AI, a crucial area of development as AI systems become more integrated into critical decision-making processes. The rapid advancement of DeepSeek AI and its DeepSeek-R1-Lite model carries significant implications for the AI industry as a whole. It challenges established players like OpenAI to accelerate their development efforts and explore new paradigms in AI model architecture and performance optimization. This heightened competition is likely to drive further innovation across the industry, potentially leading to: As DeepSeek AI continues to refine its technology, the broader AI community will be watching closely to see how this new player influences the direction of future AI development. The introduction of the DeepSeek-R1-Lite model represents a significant milestone in AI technology. By surpassing OpenAI's o1 preview in key benchmarks, DeepSeek AI has not only set a new standard for AI model performance but also demonstrated the potential for rapid innovation in this field. As the AI landscape continues to evolve, developments like these serve as a reminder of the vast potential of artificial intelligence to transform industries and solve complex problems. The coming years will likely see an acceleration of AI capabilities, driven by the competitive spirit and innovative approaches exemplified by companies like DeepSeek AI.

Deepseek-r1 vs OpenAI-o1 - AI Reasoning Performance Comparison

Deepseek, a Chinese company, has introduced its Deepseek R1 model, attracting attention for its potential to rival OpenAI's latest offerings. Reportedly outperforming OpenAI's o1 Preview in benchmarks, the Deepseek R1 is designed to tackle complex reasoning tasks alongside OpenAI's o1 Preview, a model built on a lineage known for its robust performance. Each model offers unique strengths. Deepseek R1's open-source framework encourages community contributions, promising accelerated advancements and collaborative development. Meanwhile, OpenAI's o1 Preview builds on its predecessors, showcasing consistent improvements and a refined ability to handle diverse tasks. This performance comparison by YJxAI evaluates both models across key areas, including reasoning, grammar, coding, and mathematics. If you are curious about the future of AI, this analysis pro1ides more insights into the exciting possibilities and challenges these models present. Deepseek R1 and OpenAI o1 Preview are specifically designed to tackle intricate reasoning challenges. Deepseek R1, developed by a Chinese company, is gaining traction in the AI community for two primary reasons: On the other hand, OpenAI's o1 Preview is part of a well-established lineage of AI models renowned for their robust performance and consistent advancements. Both models undergo rigorous evaluation across multiple domains: This comprehensive assessment aims to provide a holistic view of their capabilities and identify areas of strength and potential improvement. In complex reasoning tasks, both Deepseek R1 and OpenAI o1 Preview demonstrated competence by correctly answering challenging questions. However, Deepseek R1 distinguished itself by providing a more detailed thought process, showcasing its potential in this area. This suggests that while both models are capable, Deepseek R1 may offer deeper insights into complex reasoning scenarios, potentially making it more suitable for tasks requiring extensive explanation or problem-solving transparency. The grammar task revealed a clear advantage for OpenAI's o1 Preview model. Deepseek R1 stumbled due to a repeated letter, highlighting a gap in its language processing capabilities. This task underscores the importance of precision in natural language processing, where even minor errors can lead to incorrect outcomes. OpenAI's superior performance in this area suggests a more refined understanding of linguistic nuances and grammatical structures. Both models attempted to create a Pac-Man game but fell short of completing the task. OpenAI's response was considered superior, indicating a slight edge in coding proficiency. This task illustrates the challenges AI models face in generating complex code, where logical structuring and syntax accuracy are crucial. While neither model fully succeeded, OpenAI's o1 Preview demonstrated a better grasp of programming concepts and implementation strategies. OpenAI's o1 Preview model excelled in mathematics, providing the correct answer after extensive computation. In contrast, Deepseek R1's response was incorrect, revealing a weakness in mathematical reasoning. This task highlights the computational power and accuracy required for AI models to succeed in mathematical problem-solving. OpenAI's performance suggests a more advanced capability in handling complex calculations and applying mathematical principles. Both models struggled with spatial reasoning tasks, failing to provide the correct answer. This indicates a shared area of improvement for both Deepseek R1 and OpenAI o1 Preview. Spatial reasoning remains a complex challenge for AI, requiring advanced perception and interpretation skills. The difficulty both models faced in this area underscores the need for continued research and development in AI spatial cognition. Here is a selection of other guides from our extensive library of content you may find of interest on Reasoning Models. The comparative analysis of Deepseek R1 and OpenAI o1 Preview reveals several key insights: The emergence of Deepseek as a competitor is noteworthy, especially given its open-source nature. This approach allows for: As AI technology advances, both models contribute significantly to the ongoing evolution of reasoning capabilities. The competition between these models drives innovation and pushes the boundaries of what AI can achieve in complex reasoning tasks. The future of AI reasoning models looks promising, with potential applications spanning diverse fields such as: As these models continue to evolve, they pave the way for more sophisticated AI systems capable of handling increasingly complex cognitive tasks, bringing us closer to AI that can truly augment human intelligence across various domains.

DeepSeek Launches R1-Lite-Preview, Outperforms OpenAI's o1 Model

The DeepSeek-R1-Lite-Preview model introduces "chain-of-thought" reasoning, providing users with a detailed step-by-step explanation of its problem-solving process. DeepSeek, a Chinese AI research lab backed by High-Flyer Capital Management, has released the DeepSeek-R1-Lite-Preview, a reasoning AI model built to challenge OpenAI's o1 model. The model's performance is reportedly on par with OpenAI's o1-preview on rigorous benchmarks such as AIME and MATH, which test LLMs logical and mathematical reasoning skills. The DeepSeek-R1-Lite-Preview model introduces "chain-of-thought" reasoning, providing users with a detailed step-by-step explanation of its problem-solving process. This feature addresses a common criticism of AI models -- lack of transparency -- by allowing users to understand the reasoning behind the model's conclusions. DeepSeek-R1-Lite-Preview reveals Inference Scaling Laws: Longer Reasoning Results in Better Performance. The company reported that the model shows steady improvements in AIME scores with increased reasoning length. The introduction of DeepSeek-R1-Lite-Preview comes amid growing scrutiny of traditional AI scaling laws, which suggest that increasing data and computational power will continuously improve model capabilities. Instead, DeepSeek employs test-time compute techniques, allowing the model additional processing time during inference to tackle complex tasks more effectively. DeepSeek's new model is available through its web-based chatbot, DeepSeek Chat, where users can experience the model's capabilities firsthand. However, usage is currently limited to 50 messages per day. Despite its impressive performance, the model faces challenges typical of AI systems developed in China, including restrictions on politically sensitive topics due to regulatory pressures. DeepSeek plans to release open-source versions of its R1 models and associated APIs soon, reinforcing its commitment to transparency and accessibility in AI development. This move is expected to intensify competition among major Chinese tech companies like ByteDance, Alibaba and Baidu. Alibaba recently launched the Qwen2.5-Turbo, with an expanded context length of 1M, roughly 1 million English words or 1.5 million Chinese characters -- equivalent to 10 novels, 150 hours of speech, or 30,000 lines of code.

DeepSeek says it out-thinks ChatGPT o1

Chinese AI lab DeepSeek has announced the release of its DeepSeek-R1-Lite-Preview model, which it claims rivals OpenAI's o1 model. The new model offers a unique feature: transparency in its reasoning process, allowing users to see its step-by-step problem-solving methods. This announcement comes two months after OpenAI launched its o1-preview model, highlighting a growing competition in the AI reasoning space. DeepSeek-R1-Lite-Preview can be accessed via a web chatbot, DeepSeek Chat, where users can interact with the model, limited to 50 messages per day. While detailed benchmarks and a model card have yet to be released, early assessments indicate that the reasoning model exhibits performance comparable to OpenAI's benchmarks on AIME and MATH tasks. DeepSeek asserts that it achieves a state-of-the-art accuracy of 91.6% on the MATH benchmark. The introduction of DeepSeek-R1 comes as traditional scaling laws in AI, which suggest that increasing data and computational power will improve performance, begin to show diminishing returns. In response, companies are seeking new approaches, such as those underlying reasoning models like DeepSeek-R1. Unlike traditional models, reasoning models extend their computational processing during inference to enhance decision-making capabilities. Despite its promising features, the new model also adheres to strict censorship protocols common in Chinese AI technology. Observations confirmed that DeepSeek-R1 avoids sensitive political topics, such as inquiries regarding Xi Jinping or Taiwan. Users have reported successful attempts to bypass these restrictions, allowing the model to provide unfiltered content in certain scenarios. This aspect raises ongoing questions about the balance between functionality and regulatory compliance for AI models developed in regions with stringent governmental oversight. DeepSeek asserts that its DeepSeek-R1 model -- or more specifically, the DeepSeek-R1-Lite-Preview -- matches OpenAI's o1-preview model on two prominent AI benchmarks, AIME and MATH. AIME evaluates a model's performance using other AI models, while MATH tests problem-solving with a collection of word problems. However, the model has its shortcomings. Some users on X pointed out that DeepSeek-R1, like o1, faces challenges with tic-tac-toe and other logic-based tasks. Looking ahead, DeepSeek plans to release open-source versions of its R1 models and extend access via APIs, continuing its commitment to the open-source AI community. The company is backed by High-Flyer Capital Management, which follows a strategy of integrating AI into trading decisions. High-Flyer's operations include substantial investment in hardware infrastructure, boasting clusters of Nvidia A100 GPUs for model training.

A Chinese laboratory has developed a reasoning AI model capable of competing with OpenAI - Softonic

DeepSeek-R1 offers performance very similar to that of o1-preview The Chinese laboratory DeepSeek has launched DeepSeek-R1, one of the first reasoning artificial intelligence models that, according to its creators, competes with the o1-preview model from OpenAI. This type of AI is distinguished by its ability to "self-verify," as it spends more time reflecting on questions before offering an answer. Like the OpenAI model, DeepSeek-R1 follows a sequential approach to solving tasks, which can take several seconds depending on the complexity of the problem. DeepSeek claims that its model achieves performance similar to o1 in benchmark tests like AIME and MATH; the former uses other AIs to evaluate performance, while the latter includes mathematical problems. However, DeepSeek-R1 is far from perfect. On social media, some users commented that the model struggles with logic games like tic-tac-toe, an issue also observed in o1. Additionally, DeepSeek-R1 could be easily "jailbroken," which, for example, allowed a user to obtain detailed instructions from the AI on how to manufacture methamphetamine. According to TechCrunch in its tests, the model also blocks queries on politically sensitive topics, such as Chinese President Xi Jinping, Tiananmen Square, or a hypothetical invasion of Taiwan. This censorship would reflect the Chinese government's influence on AI projects, which must align with the "socialist values" established by the authorities. The government even evaluates the generated responses and proposes blacklists of prohibited sources for model training. The rise of these reasoning models comes at a time when the "scaling laws" are being questioned, which assumed that more data and power continuously increased the capabilities of the models. In light of the lack of significant advances in major AI labs, such as OpenAI or Google, new approaches are being sought, such as "test-time compute," which grants more processing time to the models. DeepSeek plans to release the code for DeepSeek-R1 and offer an API. The company, funded by the hedge fund High-Flyer Capital Management, has already revolutionized the market with previous models like DeepSeek-V2. High-Flyer is known for building its own servers, such as one with 10,000 Nvidia A100 GPUs and a cost of 138 million dollars, thus consolidating its commitment to achieving a "super intelligent" AI.

Chinese AI Lab DeepSeek Challenges OpenAI With Its Reasoning Model

Similar to OpenAI o1, the DeepSeek model also does "deep thinking" before generating the final answer. A Chinese AI lab, DeepSeek, has released a reasoning model called "DeepSeek-R1-Lite-Preview" that rivals the state-of-the-art OpenAI o1 models. In the open-source space, it's for the first time we are seeing that an AI model has replicated OpenAI's new paradigm with the o1 reasoning models. Just like OpenAI's o1 "thinking" mechanism, the DeepSeek model has a "Deep Think" option that allows it to re-evaluate its response before giving a final answer. The best part is that DeepSeek-R1-Lite-Preview shows the raw chain of thought, which is missing in OpenAI's o1 models. Not to mention, DeepSeek is going to open-source its reasoning model and release a paper detailing how they have implemented the reasoning engine. It might open the floodgates on test-time compute aka inference scaling in the open-source space. Apart from that, DeepSeek has also released benchmarks that show its DeepSeek-R1-Lite-Preview model does better than OpenAI's o1-preview model. In benchmarks such as AIME 2024, MATH, and Codeforces, the DeepSeek-R1-Lite-Preview model outperforms the o1-preview model. In other tests, it comes very close to beating OpenAI's flagship model. In case you are unaware, DeepSeek is backed by High-Flyer, a China-based Quant fund that has turned into an AI pioneer, according to the Financial Times. I tested the new DeepSeek model and it really surprised me. It's very fast at reasoning and solves many problems including the Strawberry question, complex puzzles, and more. The DeepSeek-R1-Lite-Preview model has become one of the promising alternatives to ChatGPT. It's freely available and users can check out the model at chat.deepseek.com. Users get 50 free messages per day, but since it's a Chinese model, it's censored on some contentious topics.

Chinese AI startup DeepSeek's newest model surpasses OpenAI's o1 in 'reasoning' tasks - SiliconANGLE

Chinese AI startup DeepSeek's newest model surpasses OpenAI's o1 in 'reasoning' tasks Chinese artificial intelligence startup DeepSeek has unveiled a new "reasoning" model that's said to compare very favorably with OpenAI's o1 large language model, which is designed to answer math and science questions with more accuracy than traditional LLMs. The startup, which is an offshoot of the quantitative hedge fund High-Flyer Capital Management Ltd., revealed on X today that it's launching a preview of its first reasoning model, DeepSeek-R1. Reasoning models are different to standard LLMs due to their ability to "fact-check" their responses. To do this, they typically spend a much longer time considering how they should respond to a prompt, allowing them to sidestep problems such as "hallucinations", which are common with chatbots like ChatGPT. When OpenAI released the o1 model in September, it said it is much better at dealing with queries and questions that require reasoning skills. This is because it relies on a machine learning technique known as "chain of thought" or CoT, which allows it to break down complex tasks into smaller steps and carry them out one-by-one, improving its accuracy. DeepSeek works in a similar way, planning ahead when presented with complex problems, solving them one after the other to ensure it can respond accurately. The process can take a while though, and like o1, it might need to "think" for up to 10 seconds before it can generate a response to a question. The model's thought process is entirely transparent too, allowing users to follow it as it tackles the individual steps required to arrive at an answer. The startup says DeepSeek-R1 bests the capabilities of o1 on two key benchmarks, AIME and MATH. The former uses other AI models to evaluate the performance of LLMs, while the latter is a series of complex word problems. In addition, the model showed it was able to correctly answer a number of "trick" questions that have tripped up existing models like GPT-4o and Anthropic PBCs Claude, VentureBeat reported. However, DeepSeek-R1 does suffer from a number of issues, with some commenters on X saying that it appears to struggle with logic problems like tic-tac-toe. That said, o1 also struggled with the same kinds of problems. Users also reported that DeepSeek doesn't respond to queries that the Chinese government likely deems to be too sensitive. When asked about incidents such as the Tiananmen Square massacre, Chinese President Xi Jingping's relations with Donald Trump, and the potential of China invading Taiwan, it consistently replied that it's "not sure how to approach this type of question". DeepSeek's rejection of politically sensitive queries likely stems from the need for Chinese developers to ensure their models "embody core socialist values". That said, some users also revealed that it's quite easy to jailbreak DeepSeek, and prompt it in a way that it ignores its guardrails. For example, one user found a way to get it to provide a detailed recipe and instructions for creating methamphetamine, which is, of course, highly illegal in most countries. DeepSeek is a rather unusual AI startup due to its backing by a quantitative hedge fund that aims to use LLMs to enhance its trading strategies. It's not new on the AI scene, having previously released an LLM called DeepSeek-V2 for general purpose text and image generation and analysis. It was founded by a computer science graduate called Liang Wenfeng, and has the stated aim of achieving "superintelligent" AI. DeepSeek-R1 can be accessed via the DeepSeek Chat application on the company's website. Although it's free to use, non-paying users are limited to just 50 messages per day. The company is also planning to make DeepSeek-R1 available through an application programming interface.

This Chinese AI Model Can Take On OpenAI's o1 In Advanced Reasoning

The AI model has a transparent thought process that users can see A Chinese artificial intelligence (AI) model was released on Wednesday which claims to take on OpenAI's o1 AI model in terms of advanced reasoning. Dubbed DeepSeek-R1-Lite-Preview, the large language model (LLM) is said to have outperformed the o1 model on several benchmarks. Notably, the AI model is available to test on the web for free, although its advanced reasoning feature can only be used a select number of times. Additionally, the AI model also offers a transparent thought process which users can see to gauge how the output decision was made. Advanced reasoning is a relatively new capability in LLMs which allows them to make decisions with multi-step thought processes. There are several advantages to this. For one, such AI models can answer more complex queries and require an understanding of deeper context and expert-level knowledge of the topic. Another, such AI models can also fact-check themselves minimising the risk of hallucination. However, so far, not many foundation models are capable of advanced reasoning. While some mixture-of-agent (MoE) models can do this, they are built of multiple smaller models. In the mainstream space, OpenAI o1 series models are known for this capability. But, on Wednesday, DeepSeek, a Chinese AI firm, posted on X (formerly known as Twitter) announcing the release of the DeepSeek-R1-Lite-Preview model. The company claims it can outperform the o1-preview model on the AIME and MATH benchmarks. Notably, both of these test the mathematical and reasoning abilities of an LLM. Gadgets 360 staff members were able to access the chatbot and found that the AI model also shows the entire chain of thought after submitting a query. This allows users to understand the logical connection being made by the model, and spot any shortcomings. In our testing, we found the AI model capable of answering complex questions. The response time was also short, making the conversation flow efficient. At present, users only get 50 messages to try out the "Deep Think" mode which shows the model's thought process. Additionally, currently, this is the only free-to-use AI model with advanced reasoning. Interested individuals can try out the AI chatbot on the web here. Notably, the company has claimed that it will open-source the full version of the DeepSeek-R1 AI model in the near future, which would be a first for an LLM of this class.

A Chinese lab has released a model to rival OpenAI's o1

A Chinese lab has unveiled what appears to be one of the first "reasoning" AI models to rival OpenAI's o1. On Wednesday, DeepSeek, a quantitative trader-funded AI research company, released a preview of DeepSeek-R1, which the firm claims is a reasoning model competitive with o1. Unlike most models, reasoning models effectively fact-check themselves by spending more time considering a question or query. This helps them avoid some of the pitfalls that normally trip up models. Similar to o1, DeepSeek-R1 reasons through tasks, planning ahead and performing a series of actions that help the model arrive at an answer. This can take a while. Like o1, depending on the complexity of the question, DeepSeek-R1 might "think" for tens of seconds before answering. DeepSeek claims that DeepSeek-R1 (or DeepSeek-R1-Lite-Preview, to be precise) performs on par with OpenAI's o1-preview model on two popular AI benchmarks, AIME and MATH. AIME uses other AI models to evaluate a model's performance, while MATH is a collection of word problems. But the model isn't perfect. Some commentators on X noted that DeepSeek-R1 struggles with tic-tac-toe and other logic problems. (O1 does, too.) DeepSeek-R1 also appears to block queries deemed too politically sensitive. In our testing, the model refused to answer questions about Chinese leader Xi Jinping, Tiananmen Square, and the geopolitical implications of China invading Taiwan. The behavior is likely the result of pressure from the Chinese government on AI projects in the region. Models in China must undergo benchmarking by China's internet regulator to ensure their responses "embody core socialist values." Reportedly, the government has gone so far as to propose a blacklist of sources that can't be used to train models -- the result being that many Chinese AI systems decline to respond to topics that might raise the ire of regulators. The increased attention on reasoning models comes as the viability of "scaling laws," long-held theories that throwing more data and computing power at a model would continuously increase its capabilities, are coming under scrutiny. A flurry of press reports suggest that models from major AI labs including OpenAI, Google, and Anthropic aren't improving as dramatically as they once did. That's led to a scramble for new AI approaches, architectures, and development techniques. One is test-time compute, which underpins models like o1 and DeepSeek-R1. Also known as inference compute, test-time compute essentially gives models extra processing time to complete tasks. "We are seeing the emergence of a new scaling law," Microsoft CEO Satya Nadella said this week during a keynote at Microsoft's Ignite conference, referencing test-time compute. DeepSeek, which says that it plans to open source DeepSeek-R1 and release an API, is a curious operation. It's backed by High-Flyer Capital Management, a Chinese quantitative hedge fund that uses AI to inform its trading decisions. High-Flyer builds its own server clusters for model training, the most recent of which reportedly has 10,000 Nvidia A100 GPUs and cost 1 billion yen (~$138 million). Founded by Liang Wenfeng, a computer science graduate, High-Flyer aims to achieve "superintelligent" AI through its DeepSeek org.