Honor Unveiled Tiangong AXB35 AI Workstation Family
The new compact series features advanced Nvidia, Intel, and AMD hardware for local AI tasks.
Updated on Sept. 18, 2026 in Computers
Live Poll
Would you consider using a compact, high-powered AI workstation for your personal or professional tasks?
Honor has officially introduced its Tiangong AXB35 workstation line, a collection of compact computers engineered specifically for local large language models and multi-agent AI workloads. The family includes three distinct configurations powered by Nvidia, Intel, and AMD technology.
Why it matters
The AXB35 series aims to bring server-grade AI processing power into a compact form factor for users who require local model deployment. This approach minimizes latency and data privacy concerns associated with cloud-based AI processing.
The Ultra model delivers up to 1 PFLOP of FP4 AI performance through the Nvidia RTX Spark platform within a 2-liter chassis weighing 1.5 kilograms. The system includes 128GB of unified memory and supports local LLMs with up to 300 billion parameters.
The players
Honor
Honor is a global technology brand that specializes in the production of smartphones, tablets, laptops, and various smart hardware solutions.
Nvidia
Nvidia is a multinational technology company recognized for designing graphics processing units and systems-on-a-chip for the gaming and AI industries.
Intel
Intel is a major semiconductor manufacturer that provides the central processing unit architecture for the Standard configuration of the AXB35.
AMD
AMD is a semiconductor designer known for creating processors and graphics technology used in the professional-grade Pro model of the AXB35.
The details
The Ultra model combines a Blackwell-based GPU with a Grace CPU, while the Pro and Standard tiers utilize AMD Ryzen AI Max+ 395 and Intel Core Ultra X7 358H processors respectively. All models support Windows 11 and Ubuntu, offering high-speed connectivity through Wi-Fi 7, two USB4 ports, and dual 10GbE ports.
Timeline
Honor unveiled the workstation family on September 18, 2026.
The Tech Race
This launch marks a significant shift as specialized hardware formerly reserved for data centers moves into compact, user-facing workstation formats. It positions Honor as a competitor in the growing market for local hardware capable of handling massive parameter AI models.
Professional users and developers can now run large language models locally without relying on cloud-based subscriptions or internet connectivity. The compact design and storage expansion slots allow for significant hardware customization within a minimal office footprint.
The takeaway
The move toward local AI workstations represents a strategic shift in how developers handle high-parameter workloads without cloud overhead. Users should look for future announcements regarding specific performance benchmarks in their primary OS environments.
Further reading
For more information on the latest hardware developments, visit our Computers section.
Live Poll
Would you consider using a compact, high-powered AI workstation for your personal or professional tasks?
