GLM5.2 Is a Milestone for China AI, but Still Hasn't Fit in Anthropic's Shoes

Research evidence excerpt

GLM5.2 Is a Milestone for China AI, but Still Hasn't Fit in Anthropic's Shoes

China (PRC) | Technology EquityJuneResearch17, 2026

Chart 1 - AA Intelligence RankingGLM5.2 Is a Milestone for China AI, but Still 70

60 Score

40 60 55 51 50 46 44 44 43 43 42 38 38 35 34

30 Intelligence

20Hasn't Fit in Anthropic's Shoes

10 Analysis

Artificial

US restriction on Fable 5 access is likely unsustainable. GLM 5.2’s global

Source: Artificial Analysis, Jefferies#3 ranking is a milestone as it is the 1st time a CN vendor enters top 3. . Blue = Global Model; Orange = Chinese model

Although it validates Zhipu's coding capability, Fable 5 ban is unlikely to

drive material revenue gain for Zhipu, given Claude/CN model alternatives, Chart 2 - AA Coding capability Ranking

60open-source model deployment and fast-changing leadership. We also view Index 50 62 59 56 51 50 48 48 47 46 46 43 42 38

40 coding 30

10anti-distillation feature from US vendors and compute shortage as risks to Analysis

CN model players. Artificial 0

.

US restriction on Fable 5 access appears unsustainable. After Anthropic launched Claude Source: Artificial Aanalysis, Jefferies

Mythos 5/Fable 5 on Jun 9, the Trump administration ordered it to block foreign nationals,

Chart 3 - AA Agentic capability Rankingincluding non-US employees in the US, from accessing the models on national security

70 69 67 67 67 66 66 62 57 53 47concerns. Unable to separate non-US users quickly, Anthropic temporarily disabled 35 global IndexAgentic

Analysisaccess and is tightening ID checks while negotiating with the govt to lift the ban. 9080706050403020100 81 76 74

GLM 5.2 ranked #3 globally in model intelligence, a milestone for Chinese AI. On Jun 13,

Zhipu launched GLM-5.2 via its Coding Plan, and subsequently released API access on Jun 17.