DeepSeek-V3.2 is now in release, integrating thinking directly into tool-use for the first time, improving its predecessor DeepSeek-V3.2 Experimental. The model supports tool-use in both thinking and non-thinking modes. China-based DeepSeek began disrupting the U.S. AI market in January with the debut of foundation models that rival those from Google and OpenAI that are available for free. The company released internal benchmark scores indicating its new model can compete with OpenAI’s GPT-5 in reasoning benchmarks and agentic tasks. A variation, DeepSeek-V3.2-Speciale, has been released for specialized math and is said to perform comparably to Google’s Gemini 3 Pro. Read more

Samsung has unveiled the Galaxy Z TriFold, which unfolds twice to present a seamless 10-inch screen suitable for mobile productivity and cinematic viewing. Measuring just 3.9 mm at its thinnest point, it’s a fairly heavy lift for a smartphone at 0.68 pounds (about the same as a can of soup) and measures 6.26 inches high by 2.9 inches wide and a mere half-inch thick when folded. That’s much lighter than a laptop, and provides an extra inch of height and two inches of width to a tablet of that weight, making it a viable option for business travelers. The device is expected to be available in Korea next week, starting at $2,450 for 16GB RAM/512GB storage, followed by launches in additional markets in 2026. Read more

Google has added four new features in Flow, its AI tool for storytelling, that offer more precise control over images and videos. The upgrades include generative imaging with Nano Banana Pro, doodle prompts, an object insertion/removal tool and camera motion. Flow was introduced in May and offers the ability to edit and build scenes using natural language. The improvements aim to make Flow output more polished. “In Flow, you can use images to serve as the characters, subjects and starting points for your clips” with pictures you upload or create in Flow with the new “Images” tab, according to the company. Read more

Google has signed on to provide collaborative multi-cloud networking with Amazon using its new “AWS Interconnect – multicloud,” now in preview. Amazon says the new collaboration will simplify multi-cloud networking via managed, private, on-demand access to cross-cloud connectivity. The jointly engineered solution leverages Google’s Cross-Cloud Interconnect technology. The idea is “to enable customers to easily build enterprise-grade applications that span both Google Cloud and AWS environments,” Amazon explains, pointing out that the adoption of multi-cloud applications is rapidly accelerating, driven at least in part by the rise in use of AI. Read more

Amazon has added agentic AI capabilities to Amazon Connect, the neural text-to-speech tool that provides AI-powered customer service support and analytics in real time. Connect is capable of neural text-to-speech in more than 30 languages and also delivers automated speech recognition. Leveraging advanced speech models from Nova Sonic, the Connect agents “deliver natural, human-like conversations, responding with the right pace, tone, and understanding across multiple languages and accents,” Amazon says. The company has also integrated third-party automated speech recognition and text-to-speech solutions from Deepgram and ElevenLabs with Connect. Read more