Tech giants and hardware makers shift focus to local AI deployment, empowering consumers and educational institutions with new AI tools integrated into devices and workflows, signalling a disruptive shift in the AI landscape.
Meta, OpenAI and several hardware makers used this week to push a clear message: AI is moving from cloud demos into everyday tools that run locally, work across devices and are being folded into business and education systems. Meta’s new Muse Glimmer 30B model is designed for on-device agentic use, while OpenAI and the National University of Singapore are broadening access to ChatGPT Edu and Codex across the campus. Around them, GIGABYTE, Google and Insta360 are all taking advantage of the same shift towards more capable, more practical computing.
Meta’s latest AI release is notable because it brings more of the workload on to consumer hardware. According to the company, Muse Glimmer 30B is its first Muse-family model to ship with open weights under the Apache 2.0 licence, and it is intended for local use on supported Macs and PCs with a single consumer GPU. Meta says the model can handle multi-step planning, coding, tool use, multimodal input and long-context reasoning in more than 100 languages. AMD has already announced day-zero support for compatible Ryzen AI Max+ systems and Radeon AI PRO R9700 workstations, underlining how quickly local AI is becoming a hardware arms race as well as a software one.
OpenAI and NUS are taking a different route, focusing on institutional adoption rather than consumer hardware. The expanded partnership will make ChatGPT Edu and Codex available to students, faculty and staff, with enterprise-level controls and an initial rollout that begins in selected courses before spreading further. OpenAI says the collaboration is meant to support teaching, research and administration, while giving students more chances to build AI projects of their own. The move follows an earlier tie-up with NUS’s School of Computing and sits within OpenAI’s wider Singapore strategy, which includes a local applied AI lab and a plan to create more than 200 technical roles in the country over the coming years.
Codex itself is becoming a broader platform. OpenAI says it is now available across ChatGPT plans and can assist with coding, review and shipping work in a range of programming languages. Recent reporting from TechRadar Pro suggests the tool is also being pushed further into autonomous and mobile use, with remote access on phones, longer-running tasks and additional workplace automation features. That wider direction matters for universities and businesses alike: the product is no longer just a coding helper, but part of a larger attempt to make AI agents useful across day-to-day work.
The hardware sector is responding in kind. GIGABYTE has unveiled its AORUS GeForce RTX 50 INFINITY series, with cooling changes, hidden power connectors and an option finished with wood-grain accents. It has also added support on its AI TOP ATOM platform for NVIDIA’s Nemotron 3.5 Lightning and NeMo Switchyard, bringing more local model deployment into reach for developers and businesses. Google, meanwhile, is rolling out Ask Maps, a Gemini-powered interface that turns Maps into a conversational planning tool, and has launched its Pixel 11 family alongside the new Pixel Tag. Insta360’s X6 rounds out the week with a more ambitious camera: 8K 360 capture, action-camera modes and on-device AI editing designed to reduce reliance on cloud workflows. Together, these launches suggest a common direction: AI and imaging are increasingly being built into the device itself, rather than added on as a remote service.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
