“Zhongke Shuguang has released the world’s first 64-thread mobile workstation, the N50 Pro, with significantly improved AI inference performance.”
2026-09-07 09:27Products & Apps🔥 47.3 heat score
2sources
1days unfolding
47.3heat score
3mentions
SummaryAI generated
On September 2, 2026, Changjiu officially launched the world’s first 64-thread mobile workstation, the N50 Pro. This device features a CPU with a maximum frequency of 4.3GHz per thread and an independent GPU with a main frequency of 3.3GHz (equipped with 16GB of video memory). It supports 128GB of memory and 8TB of storage. The entire unit weighs 1.69kg, with a thickness of only 16.9mm. It is equipped with a 4K 120Hz screen with a screen-to-body ratio of 93%, hidden interfaces, and supports gigabit-level data transmission. In terms of AI performance, this workstation can locally run 35B MoE models, with推理 speed reaching 3 to 5 times that of industry-leading products.
On September 2, 2026, Sugon launched the world’s first 64-thread mobile workstation, the N50 Pro. This device features a high-frequency CPU (up to 4.3GHz per core) and GPU (with a main frequency of 3.3GHz), supports 128GB of memory and 8TB of storage; it has a 93% screen-to-body ratio and a 120Hz 4K screen with ΔE＜1. The total weight of the device is 1.69kg. It adopts a hidden interface design, supports ten-gigabit data transmission, and multi-screen 4K output; its 16GB of dedicated GPU memory allows for local running of the 35B MoE model, with推理 speed 3 to 5 times faster than that of industry-leading products.
Sunway Technology confirmed the name of its world’s first 64-thread mobile workstation, the N50 Pro, and announced its specifications: the CPU’s single-thread frequency reaches 4.3GHz, the GPU’s main frequency is 3.3GHz, it supports 128GB of memory and 8TB of storage. The device features a 4K 120Hz screen with a screen-to-body ratio of 93%, a thickness of only 16.9mm, and a weight of 1.69kg. It uses a hidden interface design and offers a ten-gigabit data transmission rate. Its independent GPU has 16GB of video memory, and when running the 35B MoE model locally, the output rate reaches 50 Tokens/s, with AI inference speeds 3 to 5 times higher than those of mainstream products in the industry.