The Mountain View tech giant also has a take on a holiday AI showcase this year, and after recently releasing Gemini 2.0, the company followed it up with the "Flash Thinking Experimental" model which can deliver on "reasoning." However, for those who are well-versed in the AI space, it is known that OpenAI recently released their "reasonable" AI model known as "o1."
That being said, this will be focusing on a different experience as Google delivers its take on reasoning AI models which is significantly far from its usual LLM with Gemini, all offering a preview into AGI (artificial general intelligence).
Google's AI for Developers announced the arrival of its own reasoning model called the Gemini 2.0 Flash Thinking, and the company released it under experimental mode that is meant to deliver a new experience for users. The model now has a "thinking process" when it delivers responses to users that are currently running on its AI Studio platform for early adopters to try.
However, this is not your regular AI model as despite it being built on the Gemini 2.0 Flash which is meant for incredible speeds, the new Flash Thinking model will take a slightly longer time for its response.
According to Google DeepMind's chief scientist, Jeff Dean, the Gemini 2.0 Flash Thinking is a beneficiary of extra computing power as its increased inference time computation allows it to process massive information, including related prompts, to give the best possible answer.
However, it was OpenAI who, yet again, went ahead to deliver its "reasoning" model in the market with its introduction of the o1 model during the 12 Days of OpenAI showcase, with the latest model now available via ChatGPT Pro. The AI company also shared a preview of the o1 models last September.
On the other hand, Google's latest introduction of Gemini 2.0 Flash Thinking is still taking its footing in the market, with its base, the Gemini 2.0, facing mixed reviews in the market, the new reasoning model from the internet company needs more time to marinate and develop.
Google has made it known to all that it is now an AI-first company as it significantly focused on massive developments in the artificial intelligence space, and it also incorporated all of its products with the new advanced tech. One of the most recent experiences from Google was the 'Jarvis AI' which the company leaked and immediately deleted, an AI assistant that was made for automation.
The latest from Google's regular models was its multimodal AI that has been around for over a year now, Gemini, and it has delivered on various sizes including the Flash, Nano, and Pro versions. However, Google hit another snag last November when many reported that Gemini "snapped," particularly as the model responded to a prompt with a threat, saying "You are not special, Human."
More importantly, Google's AI Overview which was its dedicated generative AI development for its search engine also saw a setback when it first arrived faulty, giving users misleading information but has since been fixed. OpenAI may be the first to announce that it now has a reasoning model with o1, but Google also has a model of its own, with Gemini 2.0 Flash Thinking now in experimental mode.