Google Quietly Drops its Second Gen AI named Gemini 2.0 Pro Experimental

Update: It seems the original post was perhaps a bit too hasty. Some on social media have noted that the changelog has been updated and all references to Gemini 2.0 Pro Experimental have been scrubbed. This journalist can attest to the fact that the references are gone. A Google spokesperson spoke to TechCrunch, saying an “out-of-date release note” was “published in error.” The original story is below for context.

This relatively low-key introduction of Gemini 2.0 Pro Experimental-the next-awaited version to the Gemini 1.5 Pro model launched in February of last year-happens while the spotlight of tech circles shone brightly on the emergence of Chinese AI startup DeepSeek. DeepSeek’s recent models, readily available for companies to download and implement, are demonstrating performance levels that rival, and in some cases surpass, leading models developed by American tech giants and AI specialists. This development has sparked considerable discussion within Silicon Valley and even reached the highest levels of the U.S. government.

It will be an Experimental version of the Gemini 2.0 Pro, which has been opened for Gemini Advanced users starting from this Thursday, thereby becoming the new most advanced member of Google’s Gemini AI family, said the company. According to expectations, it should achieve “better factuality” and “stronger performance” with coding and mathematical tasks.

According to Google’s changelog, “Whether you’re tackling advanced coding challenges, like generating a specific program from scratch, or solving mathematical problems, like developing complex statistical models or quantum algorithms, 2.0 Pro Experimental will help you navigate even the most complex tasks with greater ease and accuracy.”.

Gemini Advanced can be accessed in the Google One AI Premium paid plan and via Google’s add-ons for the Google Workspace of Gemini.

Google states that Gemini 2.0 Pro Experimental is still in “early preview” and may display “unexpected behaviors” and introduce “errors.” Not all Gemini models may be accessible from the Gemini app; to date, Gemini 2.0 Pro Experimental does not have access to real-time information and some features of the Gemini app. This is entirely in keeping with the fact that it is an experimental version, and all of this may indicate that Google hasn’t fully adjusted the capabilities of this model.

“We believe in rapid iteration and bringing the best of Gemini to the world, and we want to give Gemini Advanced subscribers priority access to our latest AI innovations,” Google explains in the changelog. Your feedback helps us improve these models over time and learning from experimental launches informs how we release models more widely.” This focus on user feedback further underscores Google’s commitment to iterative development and the strategy of engaging its user base in the refinement process. Through its advance access program, Google lets subscribers to its Gemini Advanced services, not only garner valuable information but also promotes the sense of belonging and collective action in terms of shaping future AI technology for it. By using this kind of approach, Google can better tailor its models in accordance with how they would actually be used by people while assuring the ultimate product corresponds with the desires and expectations of users. The experimental launches work as testing grounds for Google, which is to say it’s where any imperfections found could be sorted before the widespread rollout. Thus, a much stronger and stable AI model comes through. In that regard, user feedback really acts as an imperative to push this iteration and further innovation that ensures Google’s leadership in an evolving AI field while bringing high-quality technology for the needs of users. It has been willing to experiment and learn from these early deployments, signifying its dedication to innovation and understanding the importance of continuous improvement in the field of artificial intelligence.