Recently, a mysterious "+" button has been added to Google Search, allowing users to upload images and documents, which can then be analyzed in depth using Google's Gemini AI model. Although Google has not released an official blog post, this feature may still be in the experimental stage and is only available to a select group of users.

Users just need to upload an image or document to activate the Google AI mode. Gemini will parse the content, allowing users to ask relevant questions and get immediate, context-aware queries. For example, after uploading a user manual, you can ask where to buy parts, or after uploading a guitar image, you can ask about the type of pickups used.

QQ20251216-140631.png

AI Arms Race Intensifies: Google Leads with Scale and In-House Chips

Google's enhanced AI features in its search mark the company's full commitment to the AI field to compete with OpenAI, the developer of ChatGPT.

Although both giants have been locked in a fierce battle over features, it has been reported that Google's recently released Gemini3 model has put OpenAI on "red alert," prompting OpenAI to release its GPT-5.2 model to maintain competitiveness.

Analysts point out that Google's scale advantage gives it significant competitiveness. Compared to OpenAI, Google can offer more features at a lower cost, such as additional Google Drive storage space for subscription users, and a larger context window when using Gemini AI tools.

Notably, the training platform for Gemini3 uses Google's own Tensor Processing Unit (TPU) chips, rather than NVIDIA chips. This means Google does not need to rely on third-party chip manufacturers to build high-performance AI models, and some observers even believe that Google has already far surpassed