Google Unveils Gemini 2.0 Flash Thinking: A Leap Forward in AI Reasoning and Transparency

Tech Rewind: 5 Biggest AI Advancements by Google in 2024

In December 2024, Google unveiled Gemini 2.0, marking a new chapter in AI capabilities. This advanced model introduces agentic features, enabling AI to perform complex tasks with autonomy and efficiency. Gemini 2.0's multimodal reasoning allows it to process and generate text, images, and audio seamlessly, enhancing user interactions across various platforms. Its integration into Google's ecosystem, including Search, Maps, and Workspace, provides users with a unified and intuitive AI experience.

Google Gemini 2.0: Redefining AI with Flash Thinking

Google has launched the highly-anticipated Gemini 2.0 with best-in-class AI features that promise a user experience enhancement. Building upon its predecessor's foundational approach, Gemini 2.0 is capable of delivering unprecedented improvement in speed and multimodal processing, agentic AI capabilities, and many more. Let's explore Gemini 2.0 features, the latest Google AI updates it offers, and how it is set to reshape the landscape of AI technology in detail.

Google Gemini 2.0 Flash Thinking : The AI That Thinks Like a Human

Google has unveiled Gemini 2.0 Flash Thinking, an experimental AI model that represents a significant advancement in reasoning and operational transparency. This innovative model introduces a feature called "Chain of Thought" reasoning, allowing users to follow its step-by-step decision-making process. By supporting multimodal inputs, including text and images, Gemini 2.0 is designed for a wide range of users, from developers to non-technical audiences. It is freely accessible through AI Studio, making sure widespread availability without financial barriers. Have you ever wondered how artificial intelligence actually "thinks"? AI decisions can often feel like a black box, making it hard to connect with the reasoning behind them. But what if you could see the step-by-step logic leading to an AI's conclusions? That's what Google's latest innovation, Gemini 2.0 Flash Thinking, aims to deliver. By introducing Chain of Thought reasoning, this experimental model not only tackles complex problems but also explains its process -- like having a conversation with a smart friend who walks you through their reasoning. What makes Gemini 2.0 even more exciting is its versatility. It's free to use through AI Studio, supports both text and image inputs, and even corrects itself when it makes mistakes. In a world where AI often feels out of reach or overly technical, Gemini 2.0 stands out as a transparent and approachable tool. Gemini 2.0 stands out due to its focus on logical reasoning and transparency, offering a range of advanced features that enhance its functionality and usability. These include: These features collectively position Gemini 2.0 as a powerful tool, surpassing earlier AI models and offering a competitive alternative to other systems, such as OpenAI's 01 preview. Gemini 2.0 is designed to address tasks that require complex reasoning, making it highly versatile across a range of real-world scenarios. Its practical applications include: This adaptability ensures that Gemini 2.0 is valuable for both technical professionals and everyday users, whether for professional projects, educational purposes, or casual problem-solving. Here are additional guides from our expansive article library that you may find useful on Google Gemini AI. Gemini 2.0 integrates innovative research and development from DeepMind, incorporating advanced techniques to enhance its reasoning and usability. Some of the key technical innovations include: These advancements make Gemini 2.0 not only a powerful AI tool but also one that is accessible and user-friendly, bridging the gap between advanced research and practical applications. One of the most new aspects of Gemini 2.0 is its ability to process multimodal inputs, such as combining text and images for cross-modal reasoning. This capability opens up new possibilities in various fields, including: Looking ahead, Google has hinted at the potential expansion of the Gemini series with "Pro" and "Ultra" versions. These future iterations may incorporate additional modalities, such as audio and video, further broadening the model's capabilities. User feedback will play a crucial role in shaping these developments, making sure that the model evolves to meet real-world needs effectively. Gemini 2.0 is freely available through AI Studio, making it accessible to a wide audience without subscription fees or hidden costs. Google actively encourages users to provide feedback on the model's performance, fostering a collaborative approach to its development. By involving the community, Google aims to refine Gemini 2.0 and align its capabilities with practical applications and user expectations. This open-access approach not only provide widespread access tos advanced AI technology but also ensures that the model continues to evolve based on real-world use cases and insights from a diverse range of users.

Google Introduces Gemini 2.0 Flash Thinking for Enhanced AI Reasoning

Advanced AI Reasoning Arrives with Google's Gemini 2.0 Flash Thinking Google has released the Gemini 2.0 Flash Thinking model to introduce extra reasoning abilities to its Gemini 2.0 Flash AI series. The new model has also been developed to demonstrate its "thinking process," enhance reasoning capabilities, and compete with the existing top AI systems available in the market, such as OpenAI's O1 range. This experimental model is available in Google AI Studio and Vertex AI. It can also be integrated with other developers through the Gemini API integration. In a post on X (formerly Twitter), DeepMind's chief scientist, Jeff Dean, highlighted the model's functionalities. Dean demonstrated how the Thinking Mode breaks down a problem into smaller components, allowing the model to present its reasoning steps before delivering a solution. A demo video shown by Dean showed how the model can solve a physics problem and how it does it. As another example, Logan Kilpatrick, the product lead at Google AI Studio, presented that the model can solve mathematical problems with text and image inputs.

Google Unveils Gemini 2.0 Flash Thinking AI Model to Take On OpenAI o1

Recently, OpenAI released the full version of reasoning-focused o1 series Google released a new artificial intelligence (AI) model in the Gemini 2.0 family on Thursday which is focused on advanced reasoning. Dubbed Gemini 2.0 Thinking, the new large language model (LLM) increases the inference time to allow the model to spend more time on a problem. The Mountain View-based tech giant claims that it can solve complex reasoning, mathematics, and coding tasks. Additionally, the LLM is said to perform tasks at a higher speed, despite the increased processing time. In a post on X (formerly known as Twitter), Jeff Dean, the Chief Scientist at Google DeepMind, introduced the Gemini 2.0 Flash Thinking AI model and highlighted that the LLM is "trained to use thoughts to strengthen its reasoning." It is currently available in Google AI Studio, and developers can access it via the Gemini API. Gadgets 360 staff members were able to test the AI model and found that the advanced reasoning focused Gemini model solves complex questions that are too difficult for the 1.5 Flash model with ease. In our testing, we found the typical processing time to be between three to seven seconds, a significant improvement compared to OpenAI's o1 series which can take upwards of 10 seconds to process a query. The Gemini 2.0 Flash Thinking also shows its thought process, where users can check how the AI model reached the result and the steps it took to get there. We found that the LLM was able to find the right solution eight out of 10 times. Since it is an experimental model, the mistakes are expected. While Google did not reveal the details about the AI model's architecture, it highlighted its limitations in a developer-focused blog post. Currently, the Gemini 2.0 Flash Thinking has an input limit of 32,000 tokens. It can only accept text and images as inputs. It only supports text as output and has a limit of 8,000 tokens. Further, the API does not come with built-in tool usage such as Search or code execution.

Google unveils new reasoning model Gemini 2.0 Flash Thinking to rival OpenAI o1

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More In its latest push to redefine the AI landscape, Google has announced Gemini 2.0 Flash Thinking, a multimodal reasoning model capable of tackling complex problems with both speed and transparency. In a post on the social network X, Google CEO Sundar Pichai wrote that it was: "Our most thoughtful model yet:)" And on the developer documentation, Google explains "Thinking Mode is capable of stronger reasoning capabilities in its responses than the base Gemini 2.0 Flash model." It supports just 32,000 tokens of input (about 50-60 pages worth of text) and can produce 8,000 tokens per output response. In a side panel on Google AI Studio, the company claims it is best for "multimodal understanding, reasoning" and "coding." Full details of the model's training process, architecture, licensing, and costs have yet to be released. Right now, it shows zero cost per token in the Google AI Studio. Accessible and more transparent reasoning Unlike competitor reasoning models o1 and o1 mini from OpenAI, Gemini 2.0 enables users to access its step-by-step reasoning through a dropdown menu, offering clearer, more transparent insight into how the model arrives at its conclusions. By allowing users to see how decisions are made, Gemini 2.0 addresses long-standing concerns about AI functioning as a "black box," and brings this model -- licensing terms still unclear -- to parity with other open source models fielded by competitors. My early simple tests of the model showed it correctly and speedily (within 1-3 seconds) answered some notoriously tricky questions for other AI models, such as counting the number of letter Rs in the word "Strawberry." (See screenshot above). In another test, when comparing two decimal numbers (9.9 and 9.11), the model systematically broke the problem into smaller steps, from analyzing whole numbers to comparing decimal places. These results are back up by independent third-party analysis from LM Arena, which named Gemini 2.0 Flash Thinking the number one performing model across all LLM categories. Native support for image uploads and analysis In a further improvement over the rival OpenAI o1 family, Gemini 2.0 Flash Thinking is designed to process images from the jump. o1 launched initially as a text only model, but has since expanded to include image and file upload analysis. Both models can also only return text, at this time. Gemini 2.0 Flash Thinking also does not currently support grounding with Google Search, or integration with other Google apps and external third-party tools, according to the developer documentation. Gemini 2.0 Flash Thinking's multimodal capability expands its potential use cases, enabling it to tackle scenarios that combine different types of data. For example, in one test, the model solved a puzzle that required analyzing textual and visual elements, demonstrating its versatility in integrating and reasoning across formats. Developers can leverage these features via Google AI Studio and Vertex AI, where the model is available for experimentation. As the AI landscape grows increasingly competitive, Gemini 2.0 Flash Thinking could mark the beginning of a new era for problem-solving models. Its ability to handle diverse data types, offer visible reasoning, and perform at scale positions it as a serious contender in the reasoning AI market, rivaling OpenAI's o1 family and beyond.

Google unveils Gemini 2.0 Flash 'Thinking' model for AI reasoning

Google has released an experimental model, Gemini 2.0 Flash Thinking, designed to showcase the model's reasoning process while solving complex problems. Available on Google AI Studio, this model focuses on multimodal understanding, reasoning, and coding. It is intended to tackle challenging problems in fields such as programming, math, and physics. The Gemini 2.0 Flash Thinking model is designed to improve reasoning over difficult challenges by generating the model's thinking process during problem-solving. This approach allows it to outperform the standard Gemini 2.0 Flash model. According to Logan Kilpatrick, head of product for AI Studio, this new model is "the first step in Google's reasoning journey." Jeff Dean, Google DeepMind's chief scientist, emphasized that the model "uses thoughts to strengthen its reasoning," resulting in improved performance through increased inference time computation -- the amount of computing used to process and analyze a question. As part of the Gemini 2.0 Flash series, the Thinking model leverages the speed and performance of Flash 2.0, offering faster computation. Google's team shared several demos, particularly in physics and probability, to showcase how the model approaches questions. For instance, when solving a physics problem, the model demonstrates its reasoning by breaking down the question into manageable steps before providing a final answer. Thinking Mode is available through two primary channels: the Gemini API and Google AI Studio. The thought process generated by the model appears in different ways depending on the platform used. When using the Gemini API, the model's thoughts appear as the first element of the content generated. In Google AI Studio, the thinking process is displayed in a separate "Thoughts" panel, which can be expanded to view the model's reasoning steps. As an experimental model, Gemini 2.0 Flash Thinking has the following limitations: The Thinking Mode model is now available in Google AI Studio and can be accessed through the Gemini API and Vertex AI. Developers can begin using it today by trying out the model with Google's Colab notebook or by integrating it into their projects.

Google released yet another Gemini AI model, and this one can reason

If Google wants to fill our phones with AI, it needs to give Pixels more storage Summary Gemini 2.0 Flash Thinking Experimental is a new reasoning model released by Google. This is Google's first major foray into AI reasoning models. Google's Gemini 2.0 Flash promises more advanced image and audio capabilities than its predecessors in a lightweight form. ✕ Remove Ads There are so many artificial intelligence products out there nowadays that it can be hard to keep track of advancements each of them makes, especially when marketing buzzwords get mixed in with underlying tech. Google Gemini was first released to the masses earlier in 2024, and although it's by far not the first AI product that Google has worked on and released, it's certainly the best effort the company has put forth in the realm. Every few months, Gemini has come out with a new version, and at the end of 2024, Gemini 2.0's experimental model has been released, and with it comes better underlying performance. With it, a new experimental reasoning model has been released. Related Google's experimental Gemini 2.0 Advanced model is here, but not for everyone Your Pixel's free subscription might come in handy Posts ✕ Remove Ads Google released the Gemini 2.0 Flash Thinking Experimental model today in AI Studio, the company's AI prototyping platform. It's built off of the newly unveiled Gemini 2.0 model, and it seems to be similar to OpenAI's o1 reasoning model, according to TechCrunch. Google DeepMind chief scientist Jeff Dean says that this new experimental reasoning model is "trained to use thoughts to strengthen its reasoning," but when Kyle Wiggers of TechCrunch tested the model out, it struggled to answer the simple question of how many times the letter "R" is in the word "strawberry" (Gemini said it appeared twice). Reasons for reasoning ✕ Remove Ads Obviously, the explosion of AI products has led to companies trying to differentiate themselves from the competition, and that usually comes through their naming and marketing conventions. All AI models "reason" to some extent. Gemini 2.0 Flash is meant to work in a "flash" as a lightweight model, and AI that can reason still works off of what it already has access to. A distinct positive of reasoning models is that they fact-check themselves, essentially showing the work that your elementary school math teacher told you was necessary back in the day. That's great, but it uses a ton of extra power and takes a longer amount of time than "typical" AI models utilize. Regardless, reasoning models are here, but the jury still remains on whether it will continue to see this amount of progress in such short timeframes in the future. Gemini 2.0 Flash, which is also available on Android devices, promises more advanced image and audio capabilities in a lightweight form, as previously mentioned. It outperforms the fully featured Gemini 1.5 Pro in many key benchmarks, even with its responses being twice as fast. Google is making serious strides quickly, and it's impressive to watch. Gemini Advanced subscribers get the most advanced features, however, but you'll have to figure out if it's worth the price for your own use case. ✕ Remove Ads

Google debuts reasoning-optimized Gemini 2.0 Flash Thinking Experimental model - SiliconANGLE

Google debuts reasoning-optimized Gemini 2.0 Flash Thinking Experimental model Google LLC today released a new artificial intelligence model, Gemini 2.0 Flash Thinking Experimental, that is optimized for reasoning tasks. The company says that the algorithm can tackle problems across fields such as programming, physics and math. It's based on another Google model, Gemini 2.0 Flash, that debuted earlier this month. The latter algorithm is positioned as a midrange option that balances response speeds with output quality. In a demo video, Google showed Gemini 2.0 Flash Thinking Experimental tackling a logic puzzle that required it to analyze a photo of four billiard balls. It developed a correct answer after deducting that the photo needs to be flipped. The model developed the solution through a process that saw it try several different approaches one after another. The algorithm's release is not unexpected. Earlier this year, Bloomberg reported that Google had assigned several AI research teams to building reasoning-optimized AI models. The Information later put the number of staffers working on the project at more than 200. Google's reasoning models reportedly use an approach known as chain-of-thought reasoning to carry out processing. The technique breaks down tasks into simpler sub-steps, which can improve AI output quality. The method was first introduced by Google researchers in a 2022 paper. Chain-of-thought reasoning also powers o1, OpenAI's rival series of reasoning models. One of the LLMs in the lineup, o1-preview, successfully completed a qualifying exam for the U.S. Math Olympiad. In internal tests, it also answered a set of science questions better than a group of experts with doctorate degrees. Earlier this month, OpenAI released a paid ChatGPT plan with an upgraded version of o1-preview. The new model can answer relatively simple programming questions with 75% fewer errors than its predecessor. It's also better at solving math problems. The launch of Gemini 2.0 Flash Thinking Experimental should create more competition for o1. Google plans to offer its new model through AI Studio, a service that developers can use to access the company's Gemini series of LLMs. "Built on 2.0 Flash's speed and performance, this model is trained to use thoughts to strengthen its reasoning," stated Google chief scientist Jeff Dean. "And we see promising results when we increase inference time computation!" The company's efforts to develop neural networks that can reason previously produced a pair of AI systems called AlphaGeometry and AlphaProof. They're designed to solve geometry problems and generate mathematical proofs, respectively.

Gemini 2.0 vs OpenAI o1 : The Future of AI Problem-Solving

Google's Gemini 2.0 Flash Thinking Experimental model represents a significant milestone in the evolution of artificial intelligence (AI). Designed to excel in reasoning and problem-solving, it emphasizes structured thinking and transparency in its decision-making process. While the model showcases remarkable capabilities in specific contexts, it also encounters notable challenges, particularly in adapting to nuanced inputs and avoiding over-reliance on its training data. This exploration by Prompt Engineering provides more insights into Gemini 2.0's strengths, limitations, and its implications for the future of reasoning-focused AI systems. Imagine this: you're working through a tricky puzzle or ethical dilemma, and just when you think you've nailed it, someone changes the rules ever so slightly. Suddenly, your carefully crafted solution doesn't hold up anymore. Frustrating, right? Now, imagine an AI designed to think through these same challenges -- breaking down problems step by step, reasoning its way to logical conclusions. That's exactly what Google's Gemini 2.0 Flash Thinking Experimental model aims to do. But here's the catch: even the most advanced AI can stumble when the problem shifts, revealing just how complex true reasoning really is. In this guide by Prompt Engineering learn what makes Gemini 2.0 such a standout in the world of AI reasoning, from its impressive ability to tackle complex scenarios to its struggle with something called "misguided attention." Whether you're a tech enthusiast, a curious observer, or someone wondering how close we are to AI that can truly "think," this exploration will shed light on the model's strengths, its limitations, and what it all means for the future of intelligent systems. Let's see if Gemini 2.0 is really up to the test -- or if it's still learning to adapt to life's curveballs. Gemini 2.0 is a state-of-the-art AI model engineered to prioritize reasoning over mere pattern recognition. Unlike traditional models, it integrates multimodal capabilities, allowing it to process and analyze diverse data types such as text, images, and more. A key feature of Gemini 2.0 is its emphasis on logical consistency and transparency, offering insights into its step-by-step decision-making process. This approach not only enhances its problem-solving abilities but also fosters trust in its outputs. Currently, Gemini 2.0 leads the Chat Mod Arena leaderboard, outperforming competitors in reasoning-based tasks. By breaking down problems into manageable components, it aims to deliver clear and logical solutions. This focus on structured reasoning positions Gemini 2.0 as a promising tool for tackling complex challenges across various domains. Gemini 2.0 demonstrates several strengths that set it apart from earlier AI models. Its design and functionality emphasize clarity, logical reasoning, and precision, making it particularly effective in specific scenarios. These strengths highlight Gemini 2.0's potential to handle sophisticated reasoning tasks, particularly when the problems align closely with its training data. Its ability to provide transparent and logical solutions makes it a valuable tool in fields requiring high levels of analytical rigor. Despite its impressive capabilities, Gemini 2.0 faces several limitations that hinder its adaptability and performance in certain contexts. These challenges underscore the complexities of developing AI systems capable of nuanced reasoning. These limitations reveal a reliance on pre-existing knowledge and a lack of flexibility when faced with novel or altered scenarios. Addressing these issues will be crucial for enhancing the model's ability to adapt and reason effectively in real-world applications. Gemini 2.0's performance varies significantly depending on the nature of the task, offering valuable insights into its capabilities and areas for improvement. By examining specific examples, a clearer picture of its strengths and weaknesses emerges. These examples illustrate the model's ability to excel in structured and familiar scenarios while highlighting its struggles with tasks requiring flexibility, creativity, or nuanced reasoning. Such insights are critical for guiding future improvements in reasoning-focused AI systems. The challenges faced by Gemini 2.0 reflect broader issues in the development of reasoning-oriented AI. Its reliance on training data and difficulty adapting to modified inputs highlight the need for more robust mechanisms to handle real-world complexity. Enhancing its ability to reason with precision and adaptability will be essential for advancing AI's practical applications. On the other hand, Gemini 2.0's transparency in reasoning represents a significant step forward. By providing clear insights into its decision-making process, the model lays a foundation for building trust in AI systems. This transparency is particularly valuable in applications where understanding the rationale behind decisions is critical, such as healthcare, education, and legal analysis. Future iterations of models like Gemini 2.0 could benefit from incorporating more advanced mechanisms for handling novel scenarios and ethical dilemmas. By addressing its current limitations, these models could pave the way for more reliable and effective AI systems capable of tackling complex, nuanced problems with greater precision and adaptability.

Google's new Gemini AI model can show you that it's really "thinking"

In September, OpenAI rolled out its o1-preview for ChatGPT, a new series of models designed with strong reasoning capabilities to deliver more thoughtful answers rather than speedy answers. Since then, multiple companies have rolled out their own reasoning AI models, like the DeepSeek-R1 model and Alibaba's QwQ-32B-Preview model. Now, it's Google's turn at bat. The new reasoning AI model from Google is called Gemini 2.0 Flash Thinking, and according to Jeff Dean, Chief Scientist for Google DeepMind, it's "an experimental model that explicitly shows its thoughts." Dean goes on to say, "this model is trained to use thoughts to strengthen its reasoning," an exciting prospect for those weary of how AI is forming its answers. Because a reasoning AI model is designed to show its thoughts as it forms an answer, it's a lot easier for the AI to realize when it's made a mistake and correct itself. That said, TechCrunch tested out Gemini 2.0 Flash Thinking Experimental and got lackluster results on one question. The site asked how many R's were in the word strawberry and received an incorrect answer of "two." This is still an experimental version of the AI model, and it's entirely possible it could be more equipped to tackle complex queries compared to simple ones. In the example below, Jeff Dean shows Google's new AI model solving a physics problem and explaining its thoughts as it goes. Google product lead Logan Kilpatrick also took to X to share an exciting example of Gemini 2.0 Flash Thinking solving "a challenging puzzle involving both visual and textual clues." In the shared video, you can see the AI model's entire thought process from start to finish. Kilpatrick drives home the point that this is "still an early version" of the model, but seeing the model in action is quite impressive. Rather than solely seeing the answer an AI model lands on, you can get a full, in-depth look at how the model started to approach the question and exactly how it arrived at its final answer. The one downside? It'll take a bit longer to get your answer. Because the AI model has to write out its thoughts rather than simply "think" them, coming to a conclusion can take longer than it would for a non-reasoning AI model. In the physics problem shown by Jeff Dean above, for example, it took just over 37 seconds for the AI model to complete its thinking process. But the end result seems incredibly worth it. For someone who turns to AI to help with a coding, physics, or math problem, being able to see the model's thought process will be able to help explain where you potentially got tripped up, which concept you failed to consider, etc. There's clearly still a long way to go before reasoning AI models are where we'd like them to be (like recognizing three R's in strawberry instead of two), but the short steps companies are making feels like positive progress.

Google unveils Gemini 2.0 Flash Thinking -- its answer to OpenAI's o1

Google has just unveiled Gemini 2.0 Flash Thinking, an AI model that combines the speed of its Flash technology with the same kind of chain of thought reasoning of OpenAI's o1 model. The new model is available to everyone right now on Google's AI Studio platform, free of charge. This free version comes with a token limit of 32,767 tokens, so it's somewhat limited compared to other Gemini models, but what's remarkable is how fast it deals with complicated problems that typically stump conventional AI models. The launch demos include riddles, probability problems and other examples needing complex thought. It's not perfect, some of the examples from members of the public show it can still make mistakes, and miss key aspects of some problem prompts. However considering that this release is no more than a few hours old, it's once again an impressive demonstration of what the DeepMind AI architecture can deliver. One of the key differences with OpenAI's approach is the fact that Gemini 2.0 Flash Thinking displays its reasoning as it goes, while o1 is much more bashful and hides its thoughts away. This is an important distinction for those who need to follow along to ensure no hallucinations are happening in very long thought chains. The new model already leapt to the number one spot in the Lmarena Chatbot Arena. That is a spectacular result for a model this new, in such a short period of time. I tested Flash Thinking with a riddle cheekily lifted from the OpenAI community 50 Really Hard Riddles web page, and it managed to nail the answer in 9.2 seconds. While this is not the kind of test that will prove anything meaningful, it's a fairly decent example of how competent this model seems to be. The other main difference between Google's new reasoning model and OpenAI is the fact that Google AI Studio allows you to adjust safety settings to check for different responses. So for instance you can tweak harassment, hate, dangerous and explicit content and see how those changes affect the model's reasoning abilities. It's still too early to definitively state an opinion one way or the other on this new model. However first indications are that Google has once again stepped up to the plate and delivered a smashing vindication of its core AI technology and abilities. Those interested in trying out the new model for free can sign up on Google's AI Studio platform, and select it from the model box in the right-hand sidebar.

Google Unveils Gemini 2.0 Flash Thinking, Challenges OpenAI o1

Looks like Google just played AGI reverse card on OpenAI with the release of Gemini 2.0 Flash Thinking. This new model comes with advanced reasoning capabilities, alongside showcasing its thoughts. Logan Kilpatrick, Google's product lead, Gemini 2.0 said that it "unlocks stronger reasoning capabilities and shows its thoughts." He said that the model can "solve complex problems with Flash speeds" while displaying its internal planning process, allowing for greater transparency in AI problem-solving. The experimental model is still in its early stages, but Kilpatrick provided an example of its potential, showcasing how it can tackle a challenging puzzle involving both visual and textual clues. Developers can try the model out today in Google AI Studio and the Gemini API. "This is just the first step in our reasoning journey, excited to see what you all think!" said Kilpatrick. "Google giving away a reasoning model for free to users in AIStudio is simply to show their power. They're back," commented a user on X. "Our most thoughtful model yet:)," posted Google chief Sundar Pichai on X. "We've been *thinking* about how to improve model reasoning and explainability," said Noam Shazeer, VP Engineering, Gemini co-lead at Google. Gemini 2.0 Flash Thinking, which builds on Google's Gemini series, is set to compete with OpenAI's o1 model, known for its impressive reasoning capabilities at a level similar to PhD students in physics, chemistry, and biology. Google recently launched Gemini 2.0 Flash, which supports multimodal inputs, including images, video, and audio, as well as multimodal outputs such as natively generated images combined with text and steerable text-to-speech (TTS) multilingual audio. It can also natively call tools like Google Search, execute code, and integrate third-party, user-defined functions. This development comes against the backdrop of OpenAI releasing the full version of the o1 model as part of its 12 days of shipmas. Besides this, it also released the o1 model in the API, upgraded with function calling, structured outputs, reasoning effort controls, developer messages, and vision inputs. A few benchmarks have been suggesting that o1 is the most powerful AI model yet and even outperforms the Claude 3.5 Sonnet in coding tasks. Google now seems to be ahead in the AGI race, while OpenAI is now playing catchup. On the 11th day of '12 Days of OpenAI,' the startup announced an update to the ChatGPT desktop application for Mac. The announcement came from John Nastos and Justin Rushing of OpenAI's ChatGPT desktop team, besides the former playing the saxophone. Nastos described the native app as 'lightweight' and easy to use without disrupting ongoing tasks. A standout feature of the app is its seamless integration with various applications on the user's computer, making it easier to interact with multiple tools directly from ChatGPT. "Our desktop app can now work with apps like Xcode, Warp, Notion, Apple and ~30 more. ChatGPT can see, understand, and automate your work in other apps -- a step along the path to a more agentic ChatGPT," said OpenAI chief product officer, Kevin Weil. "We all copy and paste things into ChatGPT all the time," Rushing said. "This feature makes that way smoother by automatically pulling context from the apps you're working with, so you can focus on asking your question, and we'll handle the rest." The app's utility extends to coding tasks. Nastos demonstrated its ability to integrate with IDEs like Xcode, showcasing how ChatGPT can assist with live coding challenges. One of the app's standout features is voice interaction, enabling users to communicate directly with ChatGPT through an advanced voice mode for faster and more natural conversations. With only one day remaining in OpenAI Shipmas, everyone is eagerly anticipating what OpenAI will unveil next to wrap up 12 days of nonstop shipping. However, so far, Google has countered every move made by OpenAI. While OpenAI has been making announcements during its '12 Days of OpenAI,' Google has introduced its own series of innovations, including the quantum chip Willow, Gemini 2, the 3D world model Genie 2, the Veo 2 video generation model, Project Astra as a universal agent, Project Mariner, Google Deep Research, and Android XR for AR/VR development." On the other hand, OpenAI has unveiled several significant updates, including the improved OpenAI o1 reasoning model, a new $200-per-month ChatGPT Pro subscription, and Sora, their text-to-video AI generator. Other notable releases include ChatGPT Search for all users, a new Projects feature for organizing chats, Canvas for collaborative writing and coding, and real-time video capabilities for ChatGPT. Moreover, OpenAI launched a range of new capabilities, such as Advanced Voice Mode with a Santa Claus voice option, a 1-800 number to call ChatGPT from landline, and ChatGPT's integration with Apple Intelligence.

Peer Inside the Mind of Google's AI With This New Experimental Gemini Model

In these early days of artificial intelligence, you might find yourself staring at an answer, wondering: how on Earth did the model get to that? With Google's newest experimental model, you can pull back the curtain on how AI thinks, as it's designed to show you its thoughts. See an AI Model's Reasoning With Gemini 2.0 Flash Thinking Building on its speedy Gemini 2.0 Flash experimental model, Google has just dropped Gemini 2.0 Flash Thinking, a version that not only offers fast solutions to complex questions, but also presents its reasoning in getting to that solution. According to Google DeepMind's Jeff Dean, the new experimental model was "trained to use thoughts to strengthen its reasoning." While Gemini 2.0 Flash Experimental is available directly on the standard Gemini desktop app, Gemini 2.0 Flash Thinking Experimental is only available on Google's AI Studio. That being said, it's easy (and free) to get started on this platform. At the time of writing, you can present the model with up to 1500 requests per day without paying a dime. You don't need a subscription to Gemini Advanced to use either of these experimental models. What's the Use (Case)? According to Google, this new experimental model is most useful when you need to understand the steps an AI model takes in getting to a response, such as when you have to "tackle difficult code and math problems." I can imagine how seeing the model's thoughts could be helpful in double-checking its work. For instance, if you were unsure about a response, you can review the model's step-by-step reasoning to either feel more confident about the resolution, or find the exact point where things went wrong. Students could likely also benefit from this model. Say you have a fairly complicated physics problem. Instead of merely copying and pasting an AI-generated response, you could also review the steps involved in getting that result. That being said, there's still no way to know for sure that the answer or the reasoning is accurate. That's because these AI models are also still students, to some extent. I don't know about you, but these recent Google releases are starting to feel like 12 Days of Christmas: Gemini Edition. It feels like Google drops an experimental model or Gemini upgrade every day, and I have to wonder if OpenAI's ChatGPT Search coming from Google's job has anything to do with it. In any case, it's a good sign for progress, as every AI update leads to stronger models.

'Gemini 2.0 Flash Thinking' is Google's first reasoning model

Google today released an experimental "Gemini 2.0 Flash Thinking" model that "explicitly shows its thoughts" to solve complex problems. As the name suggests, it is built on "2.0 Flash's speed and performance." Google says it is "trained to think out loud," thus "leading to stronger reasoning performance." Competing on OpenAI's o1, Google shared several demos across physics and probability: Gemini 2.0 Flash Thinking is available in Google AI Studio (direct link) and Vertex AI today. You can click "Expand to view model thoughts" and see the reasoning occur in real-time before it provides the final answer. This is "just the first step in [Google's] reasoning journey."

Google reveals AI 'reasoning' model that 'explicitly shows its thoughts'

Google has introduced a new AI "reasoning" model capable of answering complex questions while also providing a rundown of its "thoughts," as reported earlier by TechCrunch. The model, called Gemini 2.0 Flash Thinking, is still experimental and will likely compete with OpenAI's o1 reasoning model. In a post on X, Google DeepMind chief scientist Jeff Dean says the model is "trained to use thoughts to strengthen its reasoning," and also benefits from the speed that comes along with the faster Gemini Flash 2.0 model. The demo shared by Dean shows how Gemini 2.0 Flash Thinking goes about answering a physics problem by "thinking" through a series of steps before offering a solution. This isn't necessarily "reasoning" in the way humans perform it, but it means the machine breaks down instructions into smaller tasks that can produce stronger outcomes. Another example, posted by Google product lead Logan Kilpatrick, shows the model reasoning its way through a problem that involves both visual and textual elements. "This is just the first step in our reasoning journey," Kilpatrick says. You can try out Gemini 2.0 Flash Thinking on Google's AI Studio. There have been quite a few notable updates in the AI space as of late, with Google revealing its upgraded Gemini 2.0 model earlier this month as part of the company's push into "agentic" AI. Meanwhile, OpenAI made the full version of its o1 reasoning model available to ChatGPT subscribers.

Google Drops Its First "Reasoning" Model to Take On OpenAI o1

It uses more compute resources and time to re-evaluate its response before generating the final answer. After OpenAI introduced its o1 reasoning model that takes some time to "think" before responding, Google has now finally released its own version of the thinking model. The new AI model is "Gemini 2.0 Flash Thinking" aka gemini-2.0-flash-thinking-exp-1219. It's an experimental preview model, and already available on AI Studio for testing and feedback. The Gemini 2.0 Flash Thinking model follows the new paradigm of test-time compute that OpenAI introduced in September. Basically, it allows the model to use more compute resources and time to re-evaluate its response before generating the final answer. In early research, it's seen that when AI models are given more time to "think" during inference, they perform far better than models trained on large parameters. Google has released its first thinking model with the smaller Gemini 2.0 Flash model, but it's expected that inference scaling will come to the larger Gemini 2.0 Pro model (Gemini-Exp-1206) as well. Google says Gemini 2.0 Flash Thinking can solve complex reasoning questions and difficult math and coding problems. And unlike OpenAI o1, it shows the raw thinking process of the model which is great for transparency. Not to mention, the new Thinking model can process multimodal inputs such as images, videos, and audio files. Finally, its knowledge cutoff date is August 2024. I briefly tested the Gemini 2.0 Flash Thinking model on AI Studio. It failed the popular Strawberry question on the first try, but on the next run, it got the answer right and said there are three r's in the word "Strawberry". Next, I asked it to find Indian states that don't have 'a' in their names. Again, it got the answer wrong. I think we should wait for the larger Gemini 2.0 Pro Thinking model which should deliver strong performance, and demonstrate the power of inference scaling. Meanwhile, on the LMSYS benchmark, Gemini's thinking model has topped the chart across all categories.

Google unveils new AI model with 'stronger reasoning capabilities'

The tech giant expressed optimism for its latest product but also admitted that it currently has limitations. Google has released a new experimental artificial intelligence (AI) model that it claims has "stronger reasoning capabilities" in its responses than the base Gemini 2.0 Flash model. Launched yesterday (19 December), Gemini 2.0 Flash Thinking Experimental is available on Google's AI prototyping platform, AI Studio. The announcement follows Google's launch of Gemini 2.0, its answer to OpenAI's ChatGPT, just last week, while OpenAI released a preview of its "complex reasoning" AI model, o1, back in September. Reasoning models are designed to fact-check themselves, making them more accurate, although these types of models often take longer to deliver results. According to Google, the way the model's 'thoughts' are returned depends on whether the user is using the Gemini API directly or making a request through AI Studio. Logan Kilpatrick, who leads the product for AI Studio, took to X to call Gemini 2.0 "the first step in [Google's] reasoning journey". Jeff Dean, chief scientist for Google DeepMind and Research, also claimed that the company saw "promising results" with the new model. However, Google has also acknowledged that the newly released model has a number of limitations. These include a 32k token input limit; text and image input only; an 8k token output limit; text only output; and no built-in tool usage such as Search or code execution. TechCrunch reported that it briefly tested the model and concluded that there was "certainly room for improvement". While the prospect of reasoning models seems attractive, owing to their ability to fact-check themselves, such models have also raised concerns, including the question of whether such an AI model could effectively cheat and deceive humans. Earlier this year, Dr Shweta Singh of the University of Warwick's argued that releasing such sophisticated models without proper scrutiny is "misguided". "To achieve its desired objective, the path or the strategy chosen by AI may not always necessarily be fair, or align with human values." Earlier this year, Stanford AI Index claimed that robust evaluations for large language models are "seriously lacking", and there is a lack standardisation in responsible AI reporting. Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news.

Not to be outdone by OpenAI, Google releases its own "reasoning" AI model

It's been a really busy month for Google as it apparently endeavors to outshine OpenAI with a blitz of AI releases. On Thursday, Google dropped its latest party trick: Gemini 2.0 Flash Thinking Experimental, which is a new AI model that uses runtime "reasoning" techniques similar to OpenAI's o1 to achieve "deeper thinking" on problems fed into it. The experimental model builds on Google's newly released Gemini 2.0 Flash and runs on its AI Studio platform, but early tests conducted by TechCrunch reporter Kyle Wiggers reveal accuracy issues with some basic tasks, such as incorrectly counting that the word "strawberry" contains two R's. These so-called reasoning models differ from standard AI models by incorporating feedback loops of self-checking mechanisms, similar to techniques we first saw in early 2023 with hobbyist projects like "Baby AGI." The process requires more computing time, often adding extra seconds or minutes to response times. Companies have turned to reasoning models as traditional scaling methods at training time have been showing diminishing returns.

Google releases its own 'reasoning' AI model | TechCrunch

Google has released what it's calling a new "reasoning" AI model -- but it's in the experimental stages, and from our brief testing, there's certainly room for improvement. The new model, called Gemini 2.0 Flash Thinking Experimental (a mouthful, to be sure), is available in AI Studio, Google's AI prototyping platform. A model card describes it as "best for multimodal understanding, reasoning, and coding," with the ability to "reason over the most complex problems" in fields such as programming, math, and physics. In a post on X, Logan Kilpatrick, who leads product for AI Studio, called Gemini 2.0 Flash Thinking Experimental "the first step in [Google's] reasoning journey." Jeff Dean, chief scientist for Google DeepMind, Google's AI research division, said in his own post that Gemini 2.0 Flash Thinking Experimental is "trained to use thoughts to strengthen its reasoning." "We see promising results when we increase inference time computation," Dean said, referring to the amount of computing used to "run" the model as it considers a question. Built on Google's recently announced Gemini 2.0 Flash model, Gemini 2.0 Flash Thinking Experimental appears to be similar in design to OpenAI's o1 and other so-called reasoning models. Unlike most AI, reasoning models effectively fact-check themselves, which helps them avoid some of the pitfalls that normally trip up models. As a drawback, reasoning models often take longer -- usually seconds to minutes longer -- to arrive at solutions. Given a prompt, Gemini 2.0 Flash Thinking Experimental pauses for a matter of seconds before responding, considering a number of related prompts and "explaining" its thinking along the way. After a while, the model summarizes what appears to be the best answer. Well -- that's what's supposed to happen. When I asked Gemini 2.0 Flash Thinking Experimental how many R's were in the word "strawberry," it said "two." Your mileage may vary. In the wake of the release of o1, there's been an explosion of reasoning models from rival AI labs -- not just Google. In early November, DeepSeek, an AI research company funded by quant traders, launched a preview of its first reasoning model, DeepSeek-R1. That same month, Alibaba's Qwen team unveiled what it claimed was the first "open" challenger to o1. What opened the floodgates? Well, for one, the search for novel approaches to refine generative AI. As my colleague Max Zeff recently reported, "brute force" techniques to scale up models are no longer yielding the improvements they once did. Not everyone's convinced that reasoning models are the best path forward. They tend to be expensive, for one, thanks to the large amount of computing power needed to run them. And while they've performed well on benchmarks so far, it's lot clear whether reasoning models can maintain their current rate of progress.

New AI Model by Google Explains into 'How' It Thinks

In a bid to solve the AI 'black box' problem, Google has released a new AI model called "Gemini 2.0 Flash Thinking," trained to generate the model's thinking process along with its responses. It is currently available as an experimental model in Google AI Studio and for direct use in the Gemini API. The model creates "promising results" when inference time computation is increased, Jeff Dean, chief scientist for Google's AI research team DeepMind, said. Inference time is the amount of time it takes a model to provide output predictions based on input data. Google isn't the first company to come out with a reasoning model. Earlier this year, OpenAI released its thinking model O1. Just like Google, OpenAI also found that its reasoning model performs better if it has more time to think. The company argues that this model is "trained to use thoughts in a way that leads to stronger reasoning capabilities." The model is capable of code understanding, geometry, and solving math problems, and generating questions adapted to a specific level of knowledge. For example, if a user asks the model to create questions for US advanced placement (AP) Physics exams, it first figures out the topics any AP Physics class would cover, develops a scenario where specific concepts apply, and specifies exam-relevant information (like any assumptions the model made). Further, before providing the user with a question, the model reviews both the question and its solution. Users can currently only input 32,000 tokens at a time into the model and the outputs the model gives them will have an 8000 token limit. They can only input text and images and can only get responses in the text format. One of the key challenges regulatory bodies -- across countries and sectors -- have pointed out in the context of AI is the lack of clarity around models' decisions. This makes them cautious about the accuracy and fairness of the output. To address this, governments have been looking at ways to make AI companies more transparent. For instance, earlier this year, the Indian Economic Advisory Council to the Prime Minister (EAC-PM) advised the government to open up the licensing of AI models' core algorithms for external audits. The council said that AI factsheets for system audits by external experts would also "demystify the black box nature of AI," as per a MediaNama report. However, the council limited its suggestion to creating fact sheets of coding/revision control logs, data sources, training procedures, performance metrics, and known limitations. OpenAI and Google's reasoning models go a step further, in that they give insights into the actual reasoning process of AI models. This can allow users and auditors to understand not just how companies build and train their models, but also how a model arrives at specific conclusions. As governments worldwide develop AI regulations, the ability to demonstrate transparent reasoning processes could help companies meet emerging requirements for AI explainability and accountability.

Gemini 2.0 Flash Thinking vs ChatGPT o1: OpenAI Thinks Deeper

After OpenAI introduced o1 reasoning models on ChatGPT, the whole AI industry took notice and started working on "test-time compute" aka inference scaling. The general consensus shifted from training larger models to giving more time to "think" during inference to unlock intelligence and reasoning capability. Recently, Google announced its first reasoning model called "Gemini 2.0 Flash Thinking" which just like ChatGPT o1, re-evaluates its response before generating the final answer. The idea is to allow the model to verify its answer by checking all the possible outcomes rigorously. Inference scaling has led to far better performance even on smaller models. Now that Google has joined the "test-time compute" bandwagon, let's compare it with OpenAI's o1 and o1-mini models. To make the comparison interesting, I have also included China's DeepSeek-R1-Lite-Preview model which takes a similar approach. On that note, let's check out the comparison between Gemini 2.0 Flash Thinking, ChatGPT o1, and DeepSeek R1 Lite. Let's start with the popular Strawberry question, in which AI models are asked to count the letter 'r'. In the first test, Google's Gemini 2.0 Flash Thinking stumbles and says there are two r's in the word "Strawberry". On the other hand, ChatGPT o1 and the smaller, o1-mini model, get the answer right on the first try itself. Finally, DeepSeek's reasoning model also correctly says there are three r's. Moving to another test, I asked all three models to list out names of Indian states that don't have 'a' in their names. While Gemini 2.0 Flash Thinking correctly says Sikkim, it also includes three other states with the letter 'a'. It simply fails to reason with words. As for ChatGPT o1, o1-mini, and DeepSeek, they come out with flying colors and mention Sikkim only. Next, I tried a complicated prompt crafted by Riley Goodside to check how well AI models can weave connections and come up with the right answer. Well, Gemini 2.0 Flash Thinking, o1-mini, and DeepSeek hallucinated a lot and got the answer wrong. ChatGPT o1 was the only model that correctly said "Final Fantasy VII" which is a JRPG video game. The Beatles (John, Ringo, Paul, and George) visited India, whose future leader Rajiv Gandhi married an Italian. Since both Gemini 2.0 Flash Thinking and ChatGPT o1 support image input, I uploaded an image containing a math problem, from Gemini's Cookbook. In this multimodal test, Gemini 2.0 Flash Thinking decimates the ChatGPT o1 model. Gemini correctly identifies the triangle as right-angled and deduces that the overlapping region is 1/4th of the circle. Now, it simply divides the circle's area by 4 and you get 9π/4 (radius is 3) which is 7.065. ChatGPT o1, on the other hand, incorrectly identifies the triangle as an isosceles triangle and comes to a wrong conclusion. I feel Google is ahead of the competition when it comes to multimodal queries, especially image processing. Google's Gemini 2.0 Flash Thinking model is definitely better and faster, but my initial impression is that it's not smarter than ChatGPT o1, and even the smaller, o1-mini model. In my testing so far, I found ChatGPT o1 to be much more thoughtful, and grounded in facts. To be fair to Gemini 2.0 Flash Thinking, the reasoning system has been developed on the smaller Gemini 2.0 Flash model so comparing it with the SOTA ChatGPT o1 is a bit unfair. I think we should wait for the larger Gemini 2.0 Pro Thinking model which should scale better, resulting in stronger reasoning performance. That said, Gemini 2.0 Flash Thinking's strength lies in its multimodal understanding including video, audio, and image processing. It's just superior to competing reasoning models. Apart from that, many users have found that Gemini 2.0 Flash Thinking solves a Putnam 2024 Problem and Three Gambler's Problem. Clearly, its use case is beyond just reasoning. Nevertheless, the race to solve reasoning and intelligence has just begun, and in 2025, we will see significant improvements on this front.