China's AI-Terminal Grading Standard Goes Live: First L1–L4 Test Results Released

A national yardstick for "AI terminals"

Per IThome (July 17), China's Ministry of Industry and Information Technology, the State Administration for Market Regulation and the Ministry of Commerce jointly released the GB/Z 177—2026 national standard series, "AI Terminal Intelligence Grading," on May 8. The AI Terminal Working Group under the China Computer Industry Association's AI committee subsequently published the first batch of test results based on the standard.

The grading system: L1 to L4

The standard ranks terminal intelligence in four ascending levels:

• L1 Responsive: basic responses to commands;
• L2 Tool: completing specific tasks as a tool;
• L3 Assistive: proactively assisting the user;
• L4 Collaborative: deep collaboration with the user, to be further defined in later revisions as the industry matures.

Higher levels mean a "smarter" device, and the framework spans phones, PCs, glasses, TVs and automotive cockpits.

First results and significance

The first round covered categories including mobile terminals, with products from vendors such as Huawei entering the evaluation list. Providing a unified, measurable grading scale moves "device intelligence" from vague marketing language toward comparable, certifiable engineering metrics — giving vendors, consumers, and procurement/compliance a concrete reference point. As the L4 collaborative tier is refined in future revisions, the competition in on-device AI will gain a clearer benchmark to aim at.