AuraTracer智迹闻
中文

EVENT DOSSIER

DeepSeek, Qianwen, and Zhipu have taken turns to appear; PC manufacturers have finally got their chance to use them.

2026-09-07 18:40 Products & Apps 🔥 40.2 heat score
1sources
1days unfolding
40.2heat score
5mentions
SummaryAI generated

On September 7, 2026, AMD introduced the concept of “Personal AI” in its opening speech at IFA 2026, aiming to bring AI computing back to local devices from the cloud. Jack Huynh, senior vice president of AMD, unveiled three new products: the Ryzen AI Max 400 (codename Gorgon Halo), the next-generation HP ZBook laptops, and the 96-core Threadripper Halo Station. The speech confirmed that models such as DeepSeek, Qianwen, and ZhiPu GLM-5.3-Flash can run locally. Among them, Qwen3.8-Flash-Next and GLM-5.3-Flash have 320 billion parameters each and can be loaded into unified memory. AMD noted that as the complexity of Agent tasks increases, cloud-based token consumption surges, leading to overbudgeting for enterprises. Localization solutions can significantly reduce costs and ensure privacy. Microsoft and SU…

Related eventsRELATED EVENTS
Key entitiesKEY ENTITIES
AMDDeepSeekRyzen AI Max 400Thousand QuestionsZhipu

Coverage · reports per dayLANGUAGE SPLIT

Entity relations
AMD × DeepSeek1AMD × Ryzen AI Max 4001AMD × Thousand Questions1AMD × Zhipu1DeepSeek × Ryzen AI Max…1DeepSeek × Thousand Que…1

SignalsSIGNALS

Keyword heat
  • AMD1
  • DeepSeek1
  • Thousand Questions1
  • Zhipu1
  • Ryzen AI Max 4001

All reports (1)SOURCES

爱范儿 zh 2026-09-07 18:40

DeepSeek, Qianwen, and Zhipu have all made their appearances one after another. PC manufacturers have finally gotten their chance to use them`.

At the opening speech for IFA 2026, AMD introduced the concept of “Personal AI,” which involves bringing AI computing back to local devices from the cloud. Jack Huynh, senior vice president at AMD, unveiled three new products: the Ryzen AI Max 400 (codename Gorgon Halo), the next-generation HP ZBook laptops, and the 96-core Threadripper Halo Station. During the speech, models such as DeepSeek, Qianwen, and Zhipu GLM-5.3-Flash were demonstrated to run locally. Among them, Qwen3.8-Flash-Next and GLM-5.3-Flash have 320 billion parameters each and can be loaded into unified memory. AMD noted that as the complexity of Agent tasks increases, cloud-based token consumption surges, leading to overbudgeting for businesses. Localization solutions can significantly reduce costs and ensure privacy. Microsoft has collaborated with SUSE to enhance Win…