NVIDIA says DLSS 5 model processes twice as fast and uses 20% less VRAM at 4K
NVIDIA says its DLSS 5 model processes twice as fast as September’s version and uses 20% less VRAM at 4K. Here’s what its NBA 2K27 demo shows.
Latest Posts
Latest Posts
Browse all posts from NeoTeo
13075 posts
Apple schedules HomePod mini 2 store availability for October 30
Apple scheduled U.S. HomePod mini 2 availability for Oct. 30, 2026, at $149. Siri AI is planned for a free update; surround needs four matched speaker…
Apple announces HomeView and a staggered Siri AI rollout
Apple announced HomeView, HomePod mini 2 and Apple TV 4K: see U.S. prices, scheduled dates, Siri AI timing and the four-speaker surround requirements.
Nvidia reportedly plans $2 billion in European AI deals
Bloomberg reports Huang’s $2 billion European deal plan and four French startups named for funding; Nvidia’s 2025 U.K. commitment was £2 billion.
Spotify says Siri can route audio without naming the app
Spotify says Siri in iOS 27 can route audio without the app name. Eligible devices are iPhone 15 Pro or later with Apple Intelligence enabled, outside…
Google rolls out Gemini 4 Argon globally to eligible users
Google announced global access to Gemini 4 Argon, with routes for subscribers, developers and businesses and API rates from $2 per million input token…
Apple plans Siri AI beta for Apple TV 4K (2nd generation or later)
Apple's planned Siri AI beta covers Apple TV 4K (2nd generation or later). The US is on its initial English list; setup requires an eligible iPhone or…
Google Meet rolls out Gemini routines and post-quantum encryption
Google Meet’s Gemini routines can draft Gmail follow-ups or add Tasks, while a separate post-quantum rollout began for web clients and meeting-room ha…
Sony announces ¥55,000 price for Japan-only Midnight Black PS5
Sony set Japan’s ¥55,000 tax-included price for PS5 Midnight Black. The separate drive costs ¥11,980; release was scheduled for Nov. 19, 2026.
Qwen Image models: Which one fits your workflow?
Compare Qwen Image models by task, access, sampling, and license, including how Qwen-Image-2.1 and Turbo differ and what one local test completed.
Musk says Starship Flight 15 will try to catch its upper stage
Elon Musk said SpaceX would try to catch Starship’s upper stage on Flight 15. His Oct. 10 estimate was four to six weeks, with no exact date.
JAXA schedules MMX launch for October 20, 2026
JAXA has scheduled MMX to launch at 04:41:03 JST on October 20, 2026, with a reserve period through November 7 and a Phobos sample return planned for…
IMDEA Networks study finds chatbot data sharing in web and Android tests
Tests in Spain in May 2026 found conversation-data disclosures in 6 of 9 web clients and 3 of 8 Android clients, plus tracking after cookie rejection.
Amazon says the U.S. Army hosted an Amazon Leo demo
Amazon says the U.S. Army hosted an Amazon Leo demo. The reported peak topped 400 Mbps; further mobile and tactical-network tests were planned.
ESWIN Computing began trading in Hong Kong on October 9
ESWIN Computing began Hong Kong trading on October 9, 2026. Its IPO raised HK$2,385.7 million net; the prospectus details customer and supplier concen…
JP Morgan’s reported estimate puts iPhone 18 Pro demand slightly above iPhone 17 Pro
Reported October supplier-order cuts and JP Morgan’s separate estimate use different measures; the iPhone 18 Pro demand picture remains unsettled.
ESA reports successful tests of two LUPEX spectrometer models
ESA reports two EMS-L development models assembled and tested. The instrument is designed to study the lunar exosphere on LUPEX, planned for launch no…
Meta settlement sets different payment routes for Alaska and Wyoming
The Meta settlement gives Alaska a 50/50 split and sends Wyoming’s installments to its General Fund, with a separate route for its Cambridge payment.
Google Meet’s post-quantum rollout began October 12
Google says Meet’s gradual post-quantum rollout began Oct. 12 for laptop web clients and room hardware. Updates are automatic, with mobile support pla…
Tencent’s bond offering draws $17.1 billion in reported orders
Tencent’s reported bond offering drew $17.1 billion in orders. Its Q2 results detail cash-flow costs linked to computing, AI and cloud investment.
Linus Torvalds used AI to review Linux code; tests confirmed one bug and found three more
Torvalds used LLM tools to clarify Linux’s assoc_array code. A user-space test confirmed one reported bug and found three more, all dating to 2013.
