Dean Ball says Kimi is strong at coding, but open-weight economics still hurt
koltregaskes · x · 2026-07-23
This post summarizes and reacts to Dean Ball’s argument about Moonshot’s Kimi and the economics of open-weight models.
Key points:
- Kimi is described as strong in agentic coding, roughly in the class of the best Q1 2026 models, but it may be more token-hungry than expected, reducing its cost advantage.
- On why Chinese labs keep shipping capable open-weight models, the take is split roughly 75% strategic blindness to AGI risk and 25% practical necessity.
- The argument is that Chinese labs struggle to compete in paid services, and users outside China generally would not pay for their models anyway, so open-weight releases are partly ideology and partly business reality.
- The broader thesis: if users move to free models instead of paying frontier labs, the economics of the frontier model market get harder.
More from AGI Musings
- Quantum physicist told his team 'I think AGI is here' — Forbes on what convinced him — Vailhem · 2026-09-11
- DeepMind alignment researcher signs open letter urging coordinated AI slowdown — vkrakovna · 2026-09-11
- WIRED: recursive self-improvement and rogue agent swarms spook AI researchers — nordicinst · 2026-09-11
- People Neglect Human Agency Both Ways: Exaggerated Doom and Complacent Optimism — jankulveit · 2026-09-11
- Garrison Lovely's 'Obsolete' on AI Replacing Labor Lands September 2026 with Heavyweight Blurbs — GarrisonLovely · 2026-09-11
- AI Doom Skeptics Hit Back: EA-Driven Apocalypse Talk Doesn't Reflect Most Top-Tier Researchers — GarrisonLovely · 2026-09-11