FrontierThe story, in brief

Kimi K3 open-weight model: China’s biggest AI is a bet on memory, not compute

2.8 trillion parameters. Moonshot AI just released the largest open-weight model ever—and it's a play on memory, not raw compute.

Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

Kimi K3 represents a significant shift in how Chinese AI labs approach model scaling. By prioritizing extended context (memory) over parameter efficiency, Moonshot is challenging the Western compute-first paradigm and signaling a different path to capability leadership in the open-weight space.

The key facts

6 to know
  1. Kimi K3: 2.8 trillion parameters

  2. Largest open-weight model released to date

  3. Moonshot AI (China)

  4. Launch date: July 16, 2026

  5. Design philosophy: memory/context over compute optimization

  6. Enters industry 3T class tier

Go to the source

AI Newsartificialintelligence-news.com

Publisher excerpt: Moonshot AI’s Kimi K3 open-weight model has been read almost entirely through its parameter count since it launchedon July 16. At 2.8 trillion parameters, it is the largest open-weight model released to date. Model sizes are usually grouped into rough brackets, and 2.8 trillion rounds into what the…
Read original report
Back to today's editionMore frontier news

Keep reading

Related stories

More from Frontier