Google Unveils New AI Models, Enhances Gemini's Thinking Capabilities

# Google # AI model

Google just dropped its much-anticipated new AI model, Gemini 2.0 Pro Experimental, on Wednesday. This launch is part of a bunch of other AI model releases they've been rolling out. They're also making their "reasoning" model, Gemini 2.0 Flash Thinking, available in the Gemini app. It's interesting because Google is pushing these models out while everyone's still buzzing about the cheaper AI reasoning models from the Chinese startup DeepSeek. DeepSeek's models are not just holding their own; they're actually matching or even beating the performance of top AI models from American tech giants. Plus, businesses can tap into DeepSeek's models through their API for a pretty sweet deal. Both Google and DeepSeek released AI reasoning models back in December, but DeepSeek's R1 model grabbed a lot more attention. Now, it looks like Google is trying to get its Gemini 2.0 Flash Thinking model more visibility by integrating it into the popular Gemini app. As for Gemini 2.0 Pro, it's the follow-up to the Gemini 1.5 Pro model that Google launched last February. Google claims it's now the top dog in their Gemini AI model family. There was a bit of a slip-up when Google accidentally announced the Gemini 2.0 Pro model's release in the Gemini app's changelog about a week ago. But this time, it's official. They're releasing an experimental version of the model on Wednesday through their AI development platforms, Vertex AI and Google AI Studio. Plus, subscribers to Gemini Advanced in the Gemini app can get their hands on Gemini 2.0 Pro too. According to Google, the new Gemini Pro model is a whiz at coding and tackling complex prompts. It's got a "better understanding and reasoning of world knowledge" than any of their previous models. Plus, Gemini 2.0 Pro can use tools like Google Search and even run code for users. The context window for Gemini 2.0 Pro is a whopping 2 million tokens, which means it can handle about 1.5 million words at once. To put that in perspective, it could take in all seven Harry Potter books in one go and still have room for about 400,000 more words. Google is also making the Gemini 2.0 Flash model available to everyone on Wednesday. This model was announced back in December and is now open to all Gemini app users. And to maybe stir up some excitement to compete with DeepSeek's models, Google is introducing a new, more budget-friendly AI model called Gemini 2.0 Flash-Lite. They say it outperforms their Gemini 1.5 Flash model but runs at the same price and speed. *TechCrunch has an AI-focused newsletter! Sign up here to get it in your inbox every Wednesday.*

Related article

Comments (32)