Tang Jie, the CEO of Chinese AI startup Zhipu, told Elon Musk he was wrong. Not with anger. With a simple correction on a public timeline.
Musk had guessed it would take until the first quarter of next year for China to release a model competitive with Anthropic’s Fable 5. Tang Jie replied: “won’t take that long.” That exchange, rare between a top Chinese tech executive and an American billionaire, did more than settle a timeline.
It opened a window onto something harder to measure: how close China’s AI sector believes it has gotten to the frontier. Zhipu is not a household name in the United States. Inside China’s AI research community, it is central.
The company’s latest model, GLM-5.2, has posted benchmark scores that match Anthropic’s Opus 4.7 and 4.8 — models released in April and May of this year. It has also outperformed OpenAI’s GPT-5.5.
Those are not marginal gains. They are direct comparisons against the most advanced systems America has fielded. Musk pushed back.
He said benchmarks do not capture “true usefulness.” He pointed to revenue. Anthropic, he argued, focuses on “maximizing useful intelligence,” and that focus shows up in the money it makes. Tang Jie answered that too.
“Focus is all we need,” he wrote, “especially focusing on what intelligence truly is.” The two men were not just arguing about a product launch. They were sketching out competing philosophies.
Musk sees usefulness as the ultimate test — what works, what sells, what people actually use. Tang Jie sees something else: a definition of intelligence that may not yet be fully understood, let alone measured by quarterly earnings. That philosophical split matters because the stakes are concrete.
Officials in both countries are watching this race closely. The exchange has revived debates about the AI talent war between the U.S. and China.
But the talent war is old news. What is new is the evidence that Chinese models have closed the gap on technical performance — at least as benchmarks measure it. Zhipu’s trajectory explains why.
The company has moved fast. Its GLM-5.2 did not just catch up to models released months earlier. It surpassed one of them.
That kind of leap, repeated across multiple Chinese labs in recent quarters, has shifted the conversation. The question is no longer whether China can build competitive AI.
It is whether the gap that once seemed structural — rooted in hardware access, research culture, or data scale — has essentially been closed. Musk remains skeptical. His point about revenue is not trivial.
Anthropic’s models are embedded in real products that generate real income. Chinese AI companies, including Zhipu, have not yet demonstrated that kind of commercial traction at scale.
But Tang Jie’s response suggests he believes that commercial usefulness will follow from a deeper understanding of intelligence itself — not the other way around. The argument will not be settled on social media. It will be settled in labs, in product releases, and in the decisions of customers who choose one model over another.
But the exchange itself was a signal. A Chinese CEO, unprompted, directly corrected Elon Musk on a question of timing and capability. He did not hedge.
He did not defer. He simply said it would not take that long.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Chinese AI CEO Tang Jie publicly corrects Elon Musk on model launch timeline”, “description”: “Tang Jie, the CEO of Chinese AI startup Zhipu, told Elon Musk he was wrong. Not with anger. With a simple correction on a public…”, “datePublished”: “2026-06-23T09:27:00+00:00T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Sophie King”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
