Estimating Kimi K3 Active Parameters

teortaxesTex · x · 2026-07-17

This post questions how a model's active parameter count fluctuates, referencing a rough estimate:

• The model is estimated to have around 55–70B active parameters.
• This back-of-envelope calculation is based on "61 layers, 2K expert hidden dim, 2.8T parameters / 896 experts."
• The math roughly breaks down to: 50 million parameters per expert per layer. If 16 experts are activated per token, adding attention, shared layers, and embedding, the overall active compute sits around 60B.
• The original poster is asking how this number might change subsequently and whether it will land around 15–20GB or another range.

Original post →

More from Infra

• RTK Terminal Compression Cuts Tokens but Leaves Your AI Coding Bill Unchanged — Bartaseth · 2026-09-11
• SF Compute founder: buying compute is 'an absolutely awful experience' right now — IgorCarron · 2026-09-11
• SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
• PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
• Local LLM server dilemma: 4x CMP-170HX (price up 53% in 20 days) vs Mac Studio M5 Ultra — rumboll · 2026-09-11
• llama.cpp lands Flash Attention tuning for RDNA4, big prefill gains on AMD — pmttyji · 2026-09-11

Infra channel →