Kimi K3 needs 25x more compute per query than DeepSeek V4.1, margin analysis shows

teortaxesTex · x · 2026-10-24

A detailed cost breakdown by teortaxesTex compares the serving economics of Moonshot's Kimi K3 against DeepSeek V4.1.

Key numbers:

The author is also pessimistic on K3.1 Pro's generalization: despite being the strongest Chinese model, it merely matches V4.1F on several benchmarks despite higher cost.

Related event: Analysis Claims Kimi Serving Costs Over 25x DeepSeek's(4 posts)→

Original post →

More from Infra

Infra channel →