Google Unveils its Latest AI Model Gemini Google Unveils its Latest AI Model Gemini
Google DeepMind CEO and Co-Founder Demis Hassabis and Sundar Pichai, CEO of Google both introduced Gemini, the tech giant’s highly anticipated... Google Unveils its Latest AI Model Gemini

Google DeepMind CEO and Co-Founder Demis Hassabis and Sundar Pichai, CEO of Google both introduced Gemini, the tech giant’s highly anticipated AI model that promises to reshape the landscape of artificial intelligence. According to the blog post by the company, Gemini promises to deliver cutting-edge capabilities and may put OpenAI on alert.

But what is Gemini? The blog states that the model is the result of extensive collaboration among teams at Google. Its multimodal design allows it to seamlessly comprehend and operate across various types of information, including text, code, audio, image, and video.

The model is available in three optimized configurations:

  • Gemini Ultra: Engineered for highly complex tasks, Gemini Ultra stands as the most capable model in the Gemini lineup.
  • Gemini Pro: Ideal for scaling across a broad spectrum of tasks, Gemini Pro offers top-notch performance and flexibility.
  • Gemini Nano: Tailored for on-device tasks, Gemini Nano is a highly efficient model, ensuring optimal performance in various applications.

In-Person and Virtual Conference

September 5th to 6th, 2024 – London

Featuring 200 hours of content, 90 thought leaders and experts, and 40+ workshops and training sessions, Europe 2024 will keep you up-to-date with the latest topics and tools in everything from machine learning to generative AI and more.


Now, what about performance? The Google team tested, using Gemini Ultra, and it exceeded human experts on the Massive Multitask Language Understanding (MMLU) benchmark, scoring an impressive 90.0%.

Additionally, Gemini Ultra achieves a state-of-the-art score of 59.4% on the new Multimodal Multitask Understanding (MMMU) benchmark, showcasing its advanced reasoning abilities across diverse domains. There’s a great video below about its ability to reason math and physics:

But according to the Google team, what sets Gemini apart is its natively multimodal design. The design is a departure from the conventional approach of stitching together separate components for different modalities. This allows Gemini to seamlessly understand and reason about inputs, outperforming existing multimodal models in nearly every domain.

The model’s multimodal reasoning abilities make it adept at making sense of complex written and visual information, while its prowess in generating high-quality code in popular programming languages positions it as a leading foundation model for coding tasks worldwide.

The reliability, scalability, and efficiency of Gemini are further enhanced by its training on Google’s AI-optimized infrastructure using Tensor Processing Units (TPUs) v4 and v5e. The introduction of Cloud TPU v5p, the most powerful TPU system to date, will expedite Gemini’s development, allowing developers and enterprise customers to train large-scale AI models faster and more efficiently.

Google also emphasized safety and responsibility as part of Gemini’s core development. In the blog, they explained that the model is undergoing the most comprehensive safety evaluations of any Google AI model thus far. Their goal is to address potential risks, including bias and toxicity, and have engaged external experts to provide diverse perspectives in evaluating the model.

Now, of course, everyone is wondering when will they be able to use Gemini. According to the team, Gemini 1.0 is rolling out across various products and platforms, starting with Gemini Pro in Google products; this includes Bard.

From there, users can expect a significantly enhanced experience, with Gemini set to be integrated into more Google services, including Search, Ads, Chrome, and Duet AI in the coming months. Developers and enterprise customers can access Gemini Pro via the Gemini API in Google AI Studio or Google Cloud Vertex AI, starting December 13.

Android developers will also have the opportunity to harness Gemini Nano’s efficiency through AICore, available in Android 14. With the release of Gemini, it seems that Google is aiming to take the lead in the AI arms race. They provided a short introduction video about Gemini below:



ODSC gathers the attendees, presenters, and companies that are shaping the present and future of data science and AI. ODSC hosts one of the largest gatherings of professional data scientists with major conferences in USA, Europe, and Asia.