LLM Digest
Subscribe

AI Storyline

20 items · 2 sources · 7 days

View as JSON

Operational story trace

Z.ai Alpha

Current stateRally cools: valuation cut follows the Ox Alpha revealstatus changed Sep 8 · no named overseas cloud provider yet

Latest change

Institutions cut Zhipu's valuation midpoint on Sep 8, and shares of both Hong Kong-listed AI model makers slid in tandem — a reversal from the 8% rally the Ox Alpha reveal triggered two weeks earlier.

Earlier contextThe story so far

After weeks of speculation about a stealth model topping community leaderboards, Z.ai confirmed "Ox Alpha" was actually GLM-5.3-Flash and shipped its weights. The model runs entirely on 100,000 domestic Chinese chips with no Nvidia hardware, at roughly 1/40 Opus 4.8's price, and by Sep 1 Z.ai was courting overseas cloud providers to host that build.

editor-curated · source-linked

State over time

● Ox Alpha unmasked as GLM-5.3-Flash, on domestic chips · Aug 26–28Valuation cut, HK shares slide · Sep 8 ●
  • Ox Alpha unmasked as GLM-5.3-Flash, on domestic chips · Aug 26–28
  • 100K-chip build, overseas CSP talks · Sep 1
  • Valuation cut, HK shares slide · Sep 8
LAUNCH · Aug 26
Z.ai confirms the mystery "Ox Alpha" model is GLM-5.3-Flash, releases weights
2 sources · watcher updated status · show sources ▾
THE TURN · Aug 27
GLM-5.3-Flash runs entirely on 100,000 domestic Chinese chips at 1/40 Opus 4.8's price — no Nvidia involved
Z.ai discloses the model behind Ox Alpha runs on a 100,000-chip domestic stack with no Nvidia hardware, at roughly 1/40 the price of Opus 4.8; Z.ai shares surge 8% on the news, while one outlet pushes back that the Chinese-chip framing is really about cost optimization, not a geopolitical statement.
9 sources · show sources ▾
AFTERMATH · Aug 28
Coverage settles on Z.ai's company background and reconfirms the Ox Alpha identity
3 sources · show sources ▾
STANDING · Aug 30–31
GLM-5.3 closes the gap on Anthropic's cyber benchmark as usage climbs and the stealth-model framing draws scrutiny
4 sources · show sources ▾
SCALE-UP · Sep 1
GLM inference build reaches 100,000 domestic chips, Z.ai eyes overseas cloud providers
1 source · watcher updated status · show source ▾
NOW · Sep 8
Institutions cut Zhipu's valuation midpoint, Hong Kong-listed AI shares slide in tandem
1 source · watcher updated status · show source ▾

What to watch — open questions

  • Which overseas cloud providers is Z.ai in talks with, and would hosting there still depend on the same domestic-chip supply?
  • Do independent evals corroborate GLM-5.3-Flash's claimed cost and benchmark advantage?
  • Is the 100,000-domestic-chip inference build reproducible at scale, or was it a one-time capacity showcase for the Ox Alpha reveal?
  • What drove the institutional valuation downgrade — GLM-5.3-Flash's own reception, broader Chinese open-weight sentiment, or an unrelated sector rotation?
How this thread was built
scout surfaced this threadeditor wrote the arc · 6 beatswatcher 3 status changes

Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.