Kimi K3 Parameters and Inference Margins Revealed
zephyr_z9 · x · 2026-07-19
The post mentions that Kimi K3 has a total of about 2.8T parameters, which scales down to roughly 1.4T when served in fp4. It is called one of the sparsest frontier models on the market, with a sparsity of only 1.7%.
It also provides an inference-side estimate: its inference business has a gross margin of at least 75%–85%. The following quote adds a perspective on DeepSeek—even if model weights and parts of the architecture are open-sourced, competitors still struggle to replicate its real cost advantages because key implementation details and operational experience remain proprietary.
Related event: Moonshot Releases 2.8T Open-Weights Model Kimi K3(14 posts)→
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