Zhipu AI hit $1B ARR in 5 months — faster than Anthropic's run to the same milestone

There’s a simple way to measure whether a Chinese AI company is catching up to American rivals: follow the money. Zhipu AI’s annualized revenue just crossed $1 billion, and it got there from $100 million in five months. Anthropic needed fifteen.

The numbers come from multiple sources cited by Chinese tech outlet Zhineng Yongxian on Friday. One investor had told the outlet earlier that Zhipu’s ARR wouldn’t hit the $1B-$1.5B range until year-end. The company’s actual pace blew past that projection by months.

Zhipu got there by betting early on a specific technical bet: coding ability. In early 2025, the company redirected resources toward improving how its models write and understand code. Founder Tang Jie described the strategy as a bet on “Coding and Reasoning” — two capabilities he argued would become essential for the agent-driven future of AI.

The bet paid off. Zhipu shipped a new flagship model roughly every two months through 2025 and into 2026, each iteration pushing coding benchmarks further. The latest, GLM-5.2, launched in June and is open-source. On Code Arena — a front-end development evaluation system with over a million users participating in blind tests — GLM-5.2 scored first among all publicly available models. Even as open-source software, it matched or beat Claude Opus 4.8 and GPT-5.5 across several core metrics.

The ARR milestone isn’t an isolated data point. Zhipu’s market capitalization on the Hong Kong Stock Exchange surpassed HK$1 trillion in late June, up more than 2,000% year-to-date.

On the revenue growth front, at least, Zhipu is moving faster than any American competitor at the same stage. Whether it can sustain that pace as it scales from $1B to $10B is the open question — the one that will determine whether this is a breakout moment or just a good run.