Google’s Gemini AI has been steadily evolving over the year, with updates that demonstrate the company’s ambitions to compete in the crowded AI market. Earlier this month, the tech giant announced Gemini 2.0 Flash. The company says this is a model designed for faster and more efficient multimodal generation tasks.
Now, Google is taking Gemini 2.0 Flash capabilities one step further with Gemini 2.0 Flash Thinking, an experimental inference model. It aims to tackle complex problems in areas such as mathematics, physics, and programming.
This multimodal inference model, now available in Google’s AI Studio, is positioned as the “first step” in what the company calls an inference journey. Logan Kilpatrick, product lead at AI Studio, described the model as an early experiment in using “thinking to enhance inference.”
Google announces new Gemini memory feature to rival ChatGPT
This will allow Gemini to remember important information about you and use it in future conversations.
He said that like most AI reasoning models, Gemini 2.0 Flash Thinking attempts to fact-check itself by breaking the problem into smaller tasks before arriving at a solution. For example, the model can provide step-by-step instructions for solving complex equations, mimicking human-like problem solving to some extent, but initial testing shows Accuracy may still decrease as it has been shown to struggle.
The update also signals increased competition between Google and OpenAI. The creators of ChatGPT released the full version of the 01 inference model just a few weeks ago and recently introduced a successor model, the o3 model, which claims to have more powerful AI inference capabilities.
OpenAI introduces o3 model family as successor to o1 model focused on AI inference
Unlike traditional AI systems that excel at narrowly defined tasks, inference models aim to mimic the adaptability of human intelligence.
However, both Google and OpenAI’s inference models share significant challenges. It requires a large amount of computing power, resulting in slow response times. For Google to really stand out in this space, it will need to deliver inference models that exceed OpenAI’s o1 capabilities at a cost significantly lower than the $200 per month price that OpenAI charges for its top-tier inference models.
However, Gemini 2.0 Flash Thinking is still in the experimental stage, so it’s too early to tell what it will cost. And since inference models often require higher operating costs, it will be interesting to see how Google prices them.
But for now, Google says it has more than 200 researchers dedicated to advancing inferential AI and is investing heavily in the field. With companies like Alibaba and DeepSeek also competing to improve inference AI, the stakes are high for Google to show that its multimodal capabilities and structured approach to problem-solving can lead to practical, scalable applications. It has become.
OpenAI announces $200 ChatGPT Pro plan
ChatGPT Pro is expected to offer all the features included in the ChatGPT Plus package, as well as exclusive access to o1 Pro mode.
December 23, 2024
Link copied!
Copy failed!
