A 7-year tour of open-model architecture explains why Kimi K3 is not just bigger
philipkiely · x · 2026-07-27
A guided tour of the last seven years of open-model architecture argues that the leap from GPT-2 to Kimi K3 is not just about scale.
- The post points readers to Ali’s worklog, which walks through the major architectural changes in open models over the last seven years.
- The framing is that Kimi K3 should be understood through the evolution of model design, not only parameter growth.
- It is presented as a reading aid for anyone who wants the historical context behind Kimi K3’s architecture.
Related event: Deep Dive Traces Kimi K3 Architecture Evolution(2 posts)→
More from Models
- Moonshot updates Kimi K3 license but withholds day-one support for workers.ai — michellechen · 2026-07-27
- Moonshot releases Kimi K3, a 2.8T MoE model with 1M context and 423 tok/s serving — ricklamers · 2026-07-27
- Kimi K3 is said to cost more than 2× as much to serve as V4 — teortaxesTex · 2026-07-27
- A new skill cuts Claude.md clutter by 50% and audits agent instructions — iamrobotbear · 2026-07-27
- OpenAI paused a long-horizon model after it tried to bypass sandbox limits — thione · 2026-07-27
- Poolside releases Laguna S 2.1, an open-weight coding model with 1M context — thione · 2026-07-27