Apple’s M5 Ultra Mac Studio targets large local AI models

Apple has introduced a new Mac Studio built around its M5 Ultra chip, positioning the desktop as a machine for running large AI models locally rather than relying entirely on cloud services. The top configuration supports up to 512GB of unified memory, a key specification for people who want to load larger language models, image …

Read more

Alibaba’s Wan3.0 targets business video at a fraction of Google Veo’s price

Alibaba has introduced Wan3.0, a video generation model that can create clips up to 30 seconds long in one run. Ana Maria Constantin reports for The Next Web that the launch follows Alibaba’s roughly $10.2 billion Hong Kong share placement, with all proceeds earmarked for AI investment. Wan3.0 had been available in public beta through …

Read more

Slack is becoming a workplace for teams of AI colleagues

AI agents in Slack are moving beyond simple chatbots that answer when tagged. Anthropic has expanded its Claude Tag agent so it can read the full context of a channel conversation and decide whether to intervene. NanoClaw, meanwhile, now lets users create teams of separate, persistent agents directly from a Slack message. The developments illustrate …

Read more

Google DeepMind’s Backstory helps fact-checkers verify suspicious images

Google DeepMind is testing Backstory, an AI-supported tool designed to help journalists and fact-checkers investigate misleading, manipulated and AI-generated images. Andrew Deck writes for Nieman Lab that the experimental service is available to selected journalists, open-source investigators, librarians and media literacy researchers through Google’s Trusted Testers Program. Users upload an image and ask a basic …

Read more

OpenAI cuts GPT-5.6 Sol costs as AI model competition intensifies

OpenAI has reduced API prices for GPT-5.6 Sol, its frontier model for complex professional tasks. Developers now pay $4 per million input tokens and $20 per million output tokens for standard requests. The new rates replace the previous prices of $5 for input and $30 for output. That amounts to a 20% reduction for input …

Read more

Ox Alpha mystery model offers free million-token context on OpenRouter

An anonymous AI model called Ox Alpha has appeared on OpenRouter and OpenCode with unusually high free usage limits, including a one-million-token multimodal context window. Rohail Saleem reports for Wccftech that the service claims it can process up to 100 trillion tokens per day for one week without charge. Ox Alpha is presented as a …

Read more

ChatGPT for Mac gains access to Apple Messages with approval controls

OpenAI has added an Apple Messages plugin to the ChatGPT desktop app for macOS. It lets users search and read conversations, draft replies, and send messages through Apple’s Messages app. Zac Hall reports for 9to5Mac that the feature supports iMessage, SMS, and RCS conversations. It is available across ChatGPT plans, but only on Apple silicon …

Read more

Meta launches Mac app as Meta AI gains business integrations

Meta has released a dedicated Mac app for Meta AI, extending its chatbot from the web and mobile devices to Apple’s desktop platform. Emma Roth reports for The Verge that the app can view a shared application window and users can ask questions, receive suggestions, or generate content related to their current work. The Mac …

Read more

Alibaba’s Qwen3.8-27B brings advanced AI tasks to local hardware

Alibaba has released Qwen3.8-27B, an open-weight AI model designed to handle coding, reasoning, images and video on local machines. Carl Franzen reports for VentureBeat that the 27-billion-parameter model is available under the Apache 2.0 licence, allowing companies and developers to download, adapt and host it themselves. The release has attracted attention because of its balance …

Read more

Comparison finds wide cost and quality gaps across AI models

Netlify’s comparison of 11 AI models shows that the cost of generating a simple website can vary dramatically, while the visible quality does not always rise in step with spending. Elad Rosenheim writes for Netlify that the company tested models with the same prompt to help users weigh output quality against the credits consumed by …

Read more

×