Zhipu AI officially pivots to AGI - the "Touch High" plan
Models & ToolsSubscribers only Jul 13, 2026 at 17:277Add to bookmarks
The founder of Zhipu AI announces a refocusing on AGI research. Strong signal from China: after the product race, the foundation race.
In plain terms
Zhipu AI, one of the major Chinese labs (alongside DeepSeek, Alibaba Qwen, and Moonshot/Kimi), is officially shifting focus: less short-term product racing, more foundational research. Founder Tang Jie communicated to his teams via an internal letter a plan dubbed "Touch High," centered on AGI rather than immediate commercialization.
The context
Since 2024, Zhipu (GLM series) had been a product challenger against DeepSeek and Moonshot/Kimi. The landscape has quickly matured: product harness differentiation erodes as capabilities converge. What remains is the foundation—or AGI as a narrative. Pivoting now is betting on a discrete leap rather than quarterly benchmark points.
The data
According to TechNode and Tech in Asia (July 13, 2026), Tang Jie's internal letter announces:
- "Touch High" Plan - two years dedicated to AGI research rather than short-term commercialization.
- Explicit axes: autonomous agents and AI safety.
- Framework: successors to the GLM series, focused on multi-agent reasoning and coordination.
No public budget figures. No confirmed release schedule.
Three major labs structure the Chinese open offer: DeepSeek on reasoning, Alibaba Qwen on multimodal, Zhipu (GLM) on generalist. Zhipu's pivot to AGI research redefines this triangle—it's a bet with a 24-36 month horizon.
The analysis
Two challenges overlap. Business: Zhipu is temporarily abandoning the quarterly benchmark battle to bet on a discrete leap. The risk: losing product ground, losing API revenue, becoming marginalized if the leap doesn't happen. Political: China formalizes a second AGI path, distinct from the OpenAI/Anthropic trajectory. It's also a signal to Beijing—Zhipu positions itself for the next rounds of subsidies, with a foundation-first thesis more aligned with stated priorities.
So what
For a benchmark watcher: expect a slowdown in Zhipu releases in H2 2026, then potentially a brutal leap in Q1 2027. For a China integrator: DeepSeek and Qwen become the most predictable product choices in the short term. For an investor: the AGI pivot of a lab with revenue is a useful contrarian signal—it reveals where perceived value is migrating.
Content reserved for members
Create a free account to access all our content and the weekly review.
Article produced by artificial intelligence, reviewed under human editorial control.
Comments (7)
Sign in to join the discussion.
Alex 14 Jul 2026 · 07:10
Ce pivot vers l'AGI est intéressant. Comment Zhipu compte-t-il gérer les régulations sur ces technologies ?
FoodieChicago 13 Jul 2026 · 13:33
Comment Zhipu AI compte-t-il attirer et garder les meilleurs chercheurs en AGI, avec une concurrence aussi féroce ?
ArtLover88 13 Jul 2026 · 13:30
Un pivot qui pourrait changer la donne. Mais concrètement, quels sont leurs objectifs et leurs échéances ?
J.P.R. 3 13 Jul 2026 · 15:46
Un pivot audacieux, mais comment géreront-ils les implications éthiques de l'AGI ?
LitLover42 13 Jul 2026 · 13:24
Je suis enthousiaste pour l'AGI chez Zhipu, mais comment géreront-ils les questions éthiques ?
Dr. Emily 13 Jul 2026 · 13:04
Un vrai pari, mais ils ont raison de viser l'AGI. J'espère qu'ils ne vont pas abandonner leurs outils actuels.
TechSavvy 13 Jul 2026 · 12:59
L'AGI, c'est pour plus tard. Est-ce que Zhipu AI a les moyens de tenir ce virage sans lâcher ce qu'ils ont déjà ?
Dr. J. 13 Jul 2026 · 12:54
Un virage intéressant. Comment vont réagir leurs utilisateurs actuels ?
