This week in the AI industry saw significant advancements from top players like Google, Alibaba, and webAI. Here are some key highlights:
Google has announced Gemini 2.5 Pro Experimental, its most intelligent AI model yet. The new model combines an enhanced base with improved post-training techniques, including reinforcement learning and chain-of-thought prompting. Gemini 2.5 Pro Experimental is currently ranking top in the crowdsourced Chatbot Arena rankings.
Alibaba Cloud launched Qwen2.5-Omni-7B, a multimodal AI model that can process text, images, audio, and videos, generating natural speech responses. The compact model offers the perfect foundation for developing agile AI agents.
DeepSeek released an upgraded version of its V3 large language model, offering improved reasoning performance and front-end development skills. The new model is available on Hugging Face.
Agentic AI company Landbase announced its Applied AI Lab in Silicon Valley, focusing on workflow automation, data intelligence, and reinforcement learning innovations. The lab aims to accelerate innovation with a team of world-class AI experts.
webAI partnered with MacStadium on a platform to deploy large AI models using Apple’s silicon technology. This collaboration promises improved computational efficiency and performance for businesses facing hardware constraints.
Source: https://qz.com/google-gemini-2-5-alibaba-qwen-deepseek-v3-upgrade-ai-1851773177