Zhipu confirms GLM-5.3 launch as Goldman raises outlook 35%
China's model race is shifting from benchmarks to balance sheets: Zhipu's founder confirmed a new flagship, GLM-5.3, is coming, and Goldman Sachs just raised the company's full-year revenue forecast by 35 percent.
Zhipu's founder has confirmed that GLM-5.3, the next flagship in the GLM-5 line, is about to launch — and the market is already pricing it as China's shot at the global frontier. Chinese media reported Friday that the founder confirmed the release, with expectations framing the upgrade as "epic-level Plus": parameter count reportedly jumping from GLM-5.2's 740 billion scale past the trillion mark, and a composite capability gain of around five points that could let it match or surpass international leaders. GLM-5.2 already scored 52 in Artificial Analysis — the first Chinese model to match US frontier models on coding ability — and sits seventh globally on front-end coding leaderboards, with DeepSeek V4 Flash close behind.
The financial signal landed the same week. Goldman Sachs raised Zhipu's full-year revenue forecast by 35 percent on the strength of its API business, lifted its 2026 annualized recurring revenue estimate for China's AI model market by 30 percent to $13 billion (from $10 billion), and pegged Zhipu's year-end ARR at $2.5 billion. The bank also raised MiniMax's 2026–2028 revenue forecasts by 22–64 percent and initiated coverage with a Buy rating. Hong Kong tech rallied on the news — Zhipu shares jumped more than 16 percent on Friday — and the upgrade lands as DeepSeek announced it will raise API prices, cracking the low-price pattern that has defined China's model wars and feeding expectations that Zhipu follows with hikes of its own. Goldman frames 2026 as the "Zhipu moment": the year Chinese labs stop selling cheap and start selling capability. Whether GLM-5.3's open weights — every GLM-5 release so far shipped MIT-licensed — back that story up is the next test.
What to watch: the actual GLM-5.3 drop, and whether it delivers the promised jump or lands closer to an incremental refresh.
If GLM-5.3 really matches the frontier on coding and agents, does the US lead in closed models still matter? Tell us in the comments.
Sources: 21st Century Business Herald (via Eastmoney) · Sohu Finance · South China Morning Post · IndexBox · Chosun Ilbo · Futunn News
