The battle for ai supremacy is intensifying. Almost exactly a week after OpenAI made its announcement, Google is today offering a preview of its next-generation Gemini 2.0 model. in a <a target="_blank" data-i13n="cpos:2;pos:1" href="https://blog.google/technology/google-deepmind/google-gemini-ai-update-december-2024/#ceo-message” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:blog post;cpos:2;pos:1;elm:context_link;itc:0;sec:content-canvas” class=”link “> Attributed to Google CEO Sundar Pichai, the company says the 2.0 is its most capable model yet, and the algorithm offers native support for image and audio output. “It will allow us to build new ai agents that bring us closer to our vision of a universal assistant,” says Pichai.
Google is doing something different with Gemini 2.0. Instead of starting today's preview by first offering its most advanced version of the model, Gemini 2.0 Pro, the search giant is starting with 2.0 Flash. Starting today, the most efficient (and affordable) model is available to all Gemini users. If you want to try it yourself, you can enable Gemini 2.0 from the drop-down menu in the Gemini web client, which will be available soon.
Going forward, Google says its main focus is on adding version 2.0 intelligence to Search (no surprise), starting with . According to the company, the new model will allow the feature to address more complex and complicated questions, including those involving multi-step math and coding problems. At the same time, following an announcement, Google plans to make ai overviews available in more languages and countries.
Looking ahead, Gemini 2.0 is already driving improvements to some of Google's most advanced ai applications, including , the multimodal ai agent that the company previewed at I/O 2024. Thanks to the new model, Google It says the latest version of Astra can converse in multiple languages and even switch between them on the fly. It can also “remember” things for longer, offers improved latency, and can access tools like Google Lens and Maps.
As expected, Gemini 2.0 Flash offers significantly better performance than its predecessor. For example, it scored 63 percent on HiddenMath, a benchmark that tests the ability of ai models to complete competitive-level math problems. By contrast, the Gemini 1.5 Flash scored 47.2 percent on that same test. But the most interesting thing here is that the experimental version of Gemini 2.0 even surpasses it in many aspects; in fact, according to data shared by Google, the only domains in which it falls behind are long context understanding and automatic speech translation.
It is for that reason that Google maintains the previous model, at least for a while longer. Along with today's announcement of Gemini 2.0, the company also introduced Deep Research, a new tool that uses the long-context capabilities of Gemini 1.5 Pro to write comprehensive reports on complicated topics.
