- Google has released a more capable "Gemini-Exp-1206" model for Gemini Advanced subscribers.
- You can start using it on Gemini Advanced by opening the drop-down menu.
- Google says it's designed to handle complex queries including coding, math, and reasoning problems.
After releasing the Gemini 2.0 Flash model for all Gemini users, Google has now unveiled a more capable, “Gemini-Exp-1206” model for paid Gemini Advanced users. It’s an experimental model, which was being tested on AI Studio for the last few weeks. The new Gemini-Exp-1206 model offers improved performance across coding, math, reasoning, and instruction following.
I have used the Gemini-Exp-1206 model on AI Studio before, and it seemed like a powerful model. It doesn’t do inference scaling like OpenAI’s o1 reasoning models, but the performance is almost there. Basically, you get a much faster model that delivers top-notch performance.
Note that Google says Gemini-Exp-1206 is an early preview so it might not work in some cases. Additionally, this model can’t access real-time information from the web and some features are currently unavailable.
If you are subscribed to Gemini Advanced, you can open the drop-down menu and select “2.0 Experimental Advanced” to use the new Gemini-Exp-1206 model. It’s available on both desktop and mobile web, but you can’t use it on the mobile app yet. And if you want to use it for free, you can head to aistudio.google.com.
Amid the ongoing “12 days of OpenAI” announcements, Google has been releasing exciting new features, projects, and models to overshadow OpenAI’s planned releases. So far, Google has launched Veo 2, its AI video generator which truly looks much more capable than competing models including Sora. Other than that, Google has demoed Project Mariner, Project Astra, and Deep Research for Advanced users.
