Google Gemini just got another big upgrade

Gemini, Google’s very own AI chatbot, has received a slew of improvements recently. And it’s only getting better. Google has just announced a bunch of new stuff for Gemini—some of it for advanced users only, though you’ll also find additions for free users.

Gemini has just launched its latest batch of updates, of which the biggest one is perhaps a new “Gemini with personalization” model. This new model is basically a further sweetened version of Gemini’s 2.0 models that allows the chatbot to analyze user search history to refine its responses. You can activate this feature by selecting “Personalization (experimental)” from the model drop-down menu within the Gemini app. And upon activation, Gemini will assess user prompts and determine if incorporating search history can enhance the response.

According to Google, some of its early testing suggests that the personalization feature is beneficial for brainstorming and generating personalized recommendations, and the company intends to keep gathering user feedback to further refine the feature’s usefulness. Examples of prompts that would benefit from this new personalization feature include inquiries about vacation destinations, YouTube content ideas, and hobby or career suggestions. If you ask Gemini about potential vacation destinations, and you’ve already recently searched how much plane tickets to Puerto Rico go for, it could suggest Puerto Rico or other Caribbean destinations among its options.

The gist of it is that Google has a rough idea of your likes and dislikes based on your search results and how you’ve interacted with them, so it can use some of that info to write more helpful responses. Of course, it won’t be good for everything—after all, we search tons of stuff, not just stuff we like. That’s why it’s a separate model and not integrated with the main Gemini 2.0 models. But it has pretty cool potential nonetheless. The company also plans to expand Gemini’s personalization capabilities in the coming months by integrating with other Google services, such as Photos and YouTube.

It’s not the only changes that are being announced. Google also says that the 2.0 Flash Thinking model, currently in an experimental phase, is now available to try for all users with additions such as file uploading as well as better performance and reasoning capabilities. This is the model that can do multi-step reasoning, like ChatGPT’s o models, to try and come up with more accurate answers even if they might take a tad longer. There’s also a version of this model that can interact with apps like YouTube and Maps. And if you’re a Gemini Advanced user who has been using this for a while, you will now have access to a 1 million token context window. A larger context window allows the AI to process and understand significantly larger amounts of information at once, leading to more nuanced and accurate results, especially for complex research tasks.

Finally, the “Gems” feature, which allows you to customize Gemini into a personalized AI expert on specific topics, is also being made available to all users at no cost. This feature was previously restricted to Gemini Advanced users. Gems are Google’s equivalent to ChatGPT’s GPTs, and they enable users to tailor the AI’s responses and expertise to their particular needs by “pre-training” a version of the model so it responds to prompts in a specific way.

These changes should now be live, but they might take a while to land for some people, so don’t be surprised if you don’t see them right now.

Source: Google