#geminiultra

Gemini Implementation: In 2023, Google introduced AI Hyper computer to train and serve generative AI models, Generative AI support in Vertex AI, Google Enterprise AI platform, Duet AI in Google Workspace, and Duet AI for Google Cloud. GPUs, TPUs, ML software and compilers, workload management, Vertex AI, and Duet AI agents in Google Workspace and Google Cloud Platform have advanced, and Google have shipped many new capabilities in their AI-optimized infrastructure.

Developer and user growth has been rapid. This year, Vertex AI’s active gen AI projects increased by almost 7X from Q2 and Q3. Forbes, Formula E, and Spotify are building agents with Vertex AI, and Anthropic, AI21 Labs, and Cohere are training them. Customer applications are astoundingly diverse and creative. Fox Sports produces more entertaining content. A Priceline digital travel concierge is being built. Six Flags is creating a digital concierge. Estée Lauder is creating a digital brand manager.

Today, Google are launching many key new AI stack features to serve Gemini, Google’s most competent and general model. It was designed to be multimodal, meaning it can generalize and seamlessly understand, operate across, and combine text, code, audio, image, and video in the same way humans see, hear, read, listen, and talk about many types of information.

Unified AI stack from Google Cloud Beginning today, Gemini is part of a vertically integrated and vertically optimized AI technology stack with several key components designed to function together:

Super-scalable AI infrastructure: Google Cloud provides super-scalable AI infrastructure for enterprises, including Google’s model training and serving infrastructure. This architecture is available as a service in Google cloud regions for Google Distributed Cloud and edge data centers. Systems-level code sign was used throughout Google AI infrastructure stack to improve training, tweaking, and serving efficiency.

World-class models: World-class models Google keep delivering AI models with different skills. Google released Pathways Language Model (Palm) in late 2022, followed by Palm 2, and now Gemini Pro. Google also introduced Med-Palm and Sec-Palm domain-specific models.

Leading enterprise AI platform for developers: Vertex AI Google AI development platform, Vertex AI, has been significantly upgraded to assist developers construct agents and integrate modern AI into their applications. Vertex AI lets companies discover, customize, augment, deploy, and manage Gemini API agents and a curated set of over 130 open-source and third-party AI models that meet Google’s business safety and quality criteria. Vertex AI uses Google Cloud’s data governance and privacy rules and provides tools to enable developers utilize models safely. Vertex AI offers Search and Conversation, low-code technologies for powerful search and multichannel conversational bots.

Duet AI Workspace and Google Cloud AI assistants: Google AI-powered colleague Duet AI helps Google Workspace and Google Cloud users. Google Workspace’s Duet AI helps users write, make graphics, analyze spreadsheets, compose and summarize emails and chat messages, and summarize meetings. Google Cloud’s Duet AI helps users create, launch, scale, and monitor apps and identify and resolve cybersecurity problems faster.

Google are happy to announce in these areas:

Supporting Google world-class infrastructure Generation AI models have increased in size and complexity, requiring more training, tuning, and inference. The demand for high-performance, scalable, and cost-efficient AI infrastructure for training and serving models is growing tremendously.

This applies to Google and their clients. YouTube, Gmail, Google Maps, Google Play, and Android have long used TPUs to train and serve AI. Gemini is trained and served on TPUs.

Google launched Cloud TPU v5p,Google’s most powerful, scalable, and flexible AI accelerator, last week. In terms of pod FLOPs, TPU v5p is 4X more scalable than v4. Cloud TPU v5e went live earlier this year. This is Google’s most cost-effective TPU ever, with 2.7X industry benchmark inference-performance-per-dollar gains over TPU v4.

Google introduced our AI Hyper computer, a revolutionary supercomputer architecture with performance-optimized hardware, open software, top ML frameworks, and flexible consumption models. AI Hyper computer has various 5th-generation TPU and NVIDIA GPU accelerator classes.

Offering Google’s latest innovative models Gemini is their most adaptable model, running efficiently on data centers and mobile devices. For very complicated operations, Gemini Ultra is Google’s largest and most capable model, Gemini Pro is their best for scaling over a wide range of workloads, and Gemini Nano is Google’s most efficient for on-device tasks. Its cutting-edge capabilities will improve AI development and scaling for developers and enterprises.

Imogen 2, Google’s most advanced text-to-image technology, was enhanced today. This version improves photorealism, text rendering, and logo generation so you can simply make text-overlapping photos and logos.

Continuing Google domain-specific model efforts with Med-Palm, we are happy to announce Meld, our healthcare industry-specific foundation models. Meld gives Vertex AI customers the strength of Google’s foundation models with medical knowledge.

Supercharging Vertex AI with Gemini The Vertex AI preview of Gemini Pro begins today. Developers can now create unique agents that process text, code, photos, and video. Vertex AI helps you deploy and manage agents to production, automatically evaluate response quality and trustworthiness, and monitor and manage them.

Vertex AI supports Gemini by discovering, customizing, augmenting, managing, and deploying agents created against the Gemini API.

Gemini agents can be customized using your own data via rapid engineering, adapter-based fine tuning like Low-Rank Adaptation (LoRA), reinforcement learning from human feedback (RLHF), and distillation. Configurable retrieval augmented generation (RAG) building blocks allow agents to retrieve, comprehend, and act on real-world information using embedding’s. Vertex AI also offers extensions to act for users in third-party apps. Grounding to improve Gemini and other AI model responses by comparing results to high-quality online and enterprise data. A variety of controls to use gen AI models safely and responsibly, including Gemini. Vertex AI now supports Gemini, and now Google announce:

Auto SxS compares models automatically. Auto SxS is faster, cheaper, and adaptable for new generative AI use cases than manual model evaluation. Vertex AI’s Model Garden now includes Mistral, Image Bind, and DITO, extending our open model ecosystem. Vertex AI Search and Conversation will soon include Gemini Pro to help you build compelling, production-grade apps quickly. Enhancing Duet AI Google goal at Duet AI is to assist customers increase productivity, competitiveness, and profitability. Duet AI for Developers and Security Operations are available now, and Gemini will be added to their offering in the coming weeks.

Duet AI for Developers speeds up coding with code completion, generation, and chat in many IDEs. It simplifies repetitive development duties like unit test generation and code explanation, faster troubleshooting and issue resolution, and reduces context-switching. Using natural language chat to ask questions speeds up skills-based learning with Duet AI.

Duet AI for Developers users can use more than 25 code-assist and knowledge-based partners’ datasets specific to their platforms to get AI assistance based on partners’ coding and data models, product documentation, best practices, and other enterprise resources.

The unified security operations platform Duet AI in Security Operations from Google Cloud can help defenders defend against cyberattacks. Gen AI can help security teams detect, investigate, and respond to threats faster. First in Chronicle, Duet AI in Security Operations lets users search vast amounts of data in seconds with custom queries generated from natural language, reduce time-consuming manual reviews, quickly surface critical context by leveraging automatic summaries of case data and alerts, and improve response time with recommendations for next steps to support incident remediation.

Google owns Duet AI’s infrastructure, foundation models, top-level integration, and user experience. Google are happy that their engineers and researchers work together to deliver Google’s latest AI technology advances with a unified product experience. Duet AI will be expanded across Big Query, Looker, Google database solutions, Apigee, and more early next year.

Fueling future AI solutions Google’s vertically integrated AI technology stack has new features, reasonable pricing makes Gemini accessible to more enterprises, and they are expanding their indemnity to shield you from copyright issues.

Developers and customers can use Gemini, Vertex AI, and Duet AI to create a powerful cloud. Google Cloud’s innovations are enabling enterprises to build, employ, and succeed with gen AI to power their digital revolutions across all industries.

Google has just unveiled Gemini, its most advanced model boasting sophisticated multimodal reasoning capabilities. Gemini is engineered for versatility, being finely optimized for three distinct sizes: Gemini Ultra, Pro, and Nano, making it compatible with a wide range of platforms, from data centers to mobile devices.

The exciting news is that Gemini is now making its way to Bard in what is Bard’s most significant upgrade to date. The integration of Gemini into Bard will take place in two phases. Starting today, Bard will incorporate a specially tailored version of Gemini Pro in English, enabling more advanced reasoning, enhanced planning, and improved comprehension. Additionally, early next year, we will introduce Bard Advanced, offering you exclusive access to our most cutting-edge models and capabilities, beginning with Gemini Ultra.

Gemini Pro In Google BARD

You can start experiencing Gemini Pro in Bard right away. Prior to its public release, we rigorously subjected Gemini Pro to a battery of industry-standard benchmarks. Impressively, Gemini Pro outperformed GPT-3.5 in six out of eight benchmarks, including the crucial MMLU (Massive Multitask Language Understanding) and GSM8K, which evaluates grade school math reasoning.

Furthermore, Gemini Pro in Bard has been finely tuned to excel in areas such as understanding, summarization, reasoning, coding, and planning. The results are remarkable, as confirmed by blind evaluations with third-party raters. Bard now stands out as the preferred free chatbot when compared to leading alternatives, a testament to its enhanced capabilities.

In an exciting collaboration, we teamed up with YouTuber and educator Mark Rober to put Bard with Gemini Pro to the ultimate test: crafting the most accurate paper airplane. Witness how Bard elevated the creative process to new heights in the video above.

Starting today, you can try out Bard with Gemini Pro for text-based prompts, with support for other modalities coming soon. Initially available in English across more than 170 countries and territories, it will expand to include additional languages and regions, including Europe, in the near future.

Gemini Ultra in an advanced version of Bard early next year

Looking ahead, be prepared for Gemini Ultra, our most expansive and capable model, engineered for tackling highly complex tasks and rapidly processing diverse forms of information, including text, images, audio, video, and code.

Your opportunity to experience Gemini Ultra will come through Bard Advanced, a cutting-edge AI experience within Bard that grants access to our premier models and capabilities. We are diligently conducting comprehensive safety checks and will soon launch a trusted tester program before opening up Bard Advanced to a broader audience early next year.

This development aligns perfectly with our steadfast commitment to responsible AI. We’ve integrated safety features into Bard based on our AI Principles, including the introduction of contextual assistance, such as Bard’s “Google it” button, which makes it easier to double-check its responses. As we continue to fine-tune Bard, your feedback will play a crucial role in our ongoing improvements.

With Gemini, we are one step closer to realizing our vision of making Bard the ultimate AI collaborator in the world. We are thrilled to keep incorporating the latest advancements into Bard and eagerly anticipate witnessing how you leverage it to create, learn, and explore. Don’t hesitate to try Bard with Gemini Pro today.

By Demis Hassabis, CEO and Co-Founder of Google DeepMind, on behalf of the Gemini team

Like with many of my research colleagues, the main focus of my life’s work has been introducing Google Gemini AI. Since I was a teenager and began creating artificial intelligence for video games, as well as during my years as a neuroscience researcher attempting to comprehend how the brain functions, I have always had the belief that if we could create computers with more intelligence, we might use them to tremendously improve humanity.

At Google DeepMind, our work is driven by the promise of a world where artificial intelligence is used ethically. We have long aspired to develop a new generation of AI models that are motivated by human perceptions and interactions with the environment. Artificial Intelligence (AI) that has the sense of a knowledgeable assistant or aid rather than a clever piece of software.

We’re a step closer to realizing this goal today with the release of Gemini, our most competent and versatile model to date. Large-scale teamwork from teams within Google, including our friends at Google Research, produced Gemini. Since it was designed from the ground up to be multimodal, it can comprehend, operate on, and combine various forms of information, such as text, code, audio, images, and videos, with ease.

Watch video Gemini: Google’s newest and most capable AI model

Google Gemini AI Types

Our most adaptable model to date, Gemini can operate well on a wide range of platforms, including mobile phones and data centers. Its cutting-edge features will greatly improve how developers and business clients use AI to create and grow. Our initial version, Gemini 1.0, has been tuned for three distinct sizes:

• Gemini Ultra: which is our largest and most powerful model for extremely complicated activities;

• Gemini Pro: which is our best model for scaling over a broad variety of jobs;

• Gemini Nano: which is our most effective model for on-device tasks.

cutting-edge performance

Our Gemini models have undergone extensive testing, and we’ve been assessing how well they perform across a broad range of jobs. Gemini Ultra outperforms the state-of-the-art findings on 30 of the 32 commonly used academic benchmarks used in large language model (LLM) research and development, from natural picture, audio, and video understanding to mathematical reasoning.

With an MMLU (massive multitask language understanding) score of 90.0%, Gemini Ultra is the first model to surpass human specialists in this domain. MMLU tests both problem-solving and general knowledge across 57 areas, including arithmetic, physics, history, law, medicine, and ethics.

Our new benchmark method to MMLU allows Gemini to make substantial gains over simply relying on its initial impression by leveraging its reasoning powers to deliberate more thoroughly before responding to challenging issues. Additionally, Gemini Ultra reaches a state-of-the-art score of 59.4% on the recently introduced MMMU benchmark, which comprises multimodal tasks demanding deliberate reasoning across multiple domains.

Gemini AI vs GPT 4

Gemini Ultra fared better than earlier state-of-the-art models using the picture benchmarks we examined, even without the use of optical character recognition (OCR) systems, which extract text from images for additional processing. These standards demonstrate Gemini’s innate multimodality and show early evidence of Gemini’s capacity for more sophisticated reasoning.

Gemini AI vs GPT-4V

Enabling global access to Gemini AI

Gemini 1.0 is currently being released on several platforms and products:

Gemini Pro for Google goods

Bard will begin using an enhanced version of Gemini Pro today in order to perform more sophisticated planning, comprehending, reasoning, and other tasks. The largest update for Bard since its release is this one. More than 170 countries and territories will be able to access it in English, and we intend to soon add support for more languages and modalities as well as new languages and places.

Additionally, Gemini is coming to Pixel. With other messaging apps arriving next year, the Pixel 8 Pro is the first smartphone designed to run Gemini Nano, the technology behind new capabilities like Summarize in the Recorder app and rolling out in Gboard’s Smart Reply, starting with WhatsApp.

Gemini will be accessible across more of our products and services, including Search, Ads, Chrome, and Duet AI, in the upcoming months.

We have already begun testing Gemini in Search, where it is improving quality and reducing latency by 40% in English in the United States, thereby speeding up our Search Generative Experience (SGE) for users.

Constructing with Gemini AI

Developers and enterprise clients can use Google Cloud Vertex AI or Google AI Studio’s Gemini API to access Gemini Pro as of December 13.

With an API key, Google AI Studio is a free online tool for developers to quickly prototype and create programs. When the time comes for a fully-managed AI platform, Vertex AI offers further capabilities from Google Cloud for enterprise security, safety, privacy, and data governance and compliance. It also enables customization of Gemini with complete data management.

Through AICore, a new system feature coming with Android 14, Android developers will also be able to create with Gemini Nano, our most effective model for on-device tasks, starting on Pixel 8 Pro devices. Register to receive an early look at AICore.

Soon to come: Gemini AI Ultra

Before releasing Gemini Ultra to the public, we’re presently finishing up thorough safety and trust assessments, which include red-teaming by dependable outside parties. We’re also fine-tuning the model and applying reinforcement learning from human feedback (RLHF) to improve its performance.

During this process, we will first make Gemini Ultra available for early testing and input to a restricted group of consumers, developers, partners, and safety and responsibility experts. Later in the year, we will roll it out to developers and enterprise customers.

We’ll also be introducing Bard Advanced early in the upcoming year, which is a state-of-the-art AI experience that allows you to access our greatest models and features, beginning with Gemini Ultra.

The Gemini AI era: opening the door to an innovative future

This marks not only the beginning of a new era for Google as we continue to responsibly and quickly expand the capabilities of our models, but also a critical milestone in the development of AI.

While Gemini has come a long way, we still have a long way to go. We’re putting a lot of effort on improving its planning and memory capacities as well as expanding the context window to handle even more data and provide better answers in future iterations.

The exciting prospects of an ethically AI-enabled society thrill us. It’s a future of innovation that will boost creativity, expand knowledge, progress research, and revolutionize the way billions of people live and work worldwide.