The case carries the radio

The socket is nineteen fifty-four

Intel’s Design-In Tools store listed an ILM kit for LGA1954 NVL-S on 30 August. Nova Lake-S will not drop into today’s LGA1851 boards.

Blindsight human trials are in the window

Elon Musk said on 24 August 2026 that Blindsight human implants are six to 12 months out. FDA Breakthrough designation is 17 September 2024. There are still zero human Blindsight implants on the public record.

VOICE is the speech trial

Neuralink’s VOICE study (NCT07224256) is recruiting at UT Southwestern. Enrollment is listed as six. Start 3 October 2025, primary completion October 2028. The speech stack is a different product from Telepathy’s cursor.

Twenty-one Neuralnauts

Neuralink says 21 people are now in human trials worldwide. The company’s own Telepathy post puts numbers on bits per second, hours a day, and a safety line of zero serious device-related events.

Apple opens Core AI so apps load a model like a font

WWDC, 8 June. The inference stack behind Apple Intelligence is now a Swift API, a .aimodel file, and a public Qwen, Mistral, and SAM 3 repository.

Apple’s bigger on-device model needs 12 GB and an iPhone 17 Pro

WWDC, 8 June. Expressive Siri and advanced dictation stay on the 12 GB phones. The base iPhone 17, at 8 GB, does not get them.

Mac Studio gets the Ultra

On 25 August Apple put M5 Max and M5 Ultra in the Studio cylinder. Pre-order now. Ships 22 September. 512 GB in late October.

Apple put 2-nanometer M6 in the Mac mini and M5 Ultra in the Studio

25 August pre-order. Ships 22 September. M6 mini from US$899. M5 Ultra Studio from US$5,499. Apple’s numbers, Apple’s tests.

Samsung’s S26 FE is first in the family onto One UI 9

Newsroom pages on 27 August put Galaxy S26 FE on Android 17 and One UI 9, on sale from 4 September in Blueberry, Graphite, and Pistachio. 17 cm FHD+, 193 g, Exynos 2500, 4,900 mAh. No price on the pages.

Z.ai and Qwen ship the same hybrid-attention graph

26 August. GLM-5.3-Flash is 320B total, 18B active, MIT. Qwen3.8-Flash-Next is 125B with 6B active. Both land on a 3:1 linear-to-sparse stack.