Zhongnong Network Releases Internal Test Version of Sugar AI Model 'AI Sugar'


The Tnkr platform aims to solve the fragmentation issue in robot development by integrating hardware, software, data, and AI models into a unified open-source ecosystem. This allows developers to collaborate on building physical robot projects, changing the traditional 'puzzle game' mode of switching between different tools.
Google Search has added a "+" button that supports uploading images or documents for in-depth analysis using the Gemini AI model. This feature is currently in an experimental phase and available to only a select group of users. After uploading, Gemini can parse the content, allowing users to ask questions instantly and get contextually relevant answers, such as asking where to purchase parts after uploading a manual.
Meta plans to release an AI model codenamed "Avocado" in the spring of 2026, which may shift to being closed-source and was trained using Alibaba's open-source model Qwen. The news has attracted market attention, causing Alibaba's stock price to rise.
Google has recently expanded its generative AI model Gemini3 on a global scale, covering 120 countries and regions. Users in supported regions can now use the Gemini3 Pro feature in Google Search AI mode.
OpenAI announced that it will terminate the API access of the GPT-4o model on February 16, 2026, and developers need to complete project adjustments within three months. The model is still available as an option for ChatGPT individual and paid users, but has been considered a legacy system.