Model Spend Arena2ND ED.
1027B · dense · 2026-08-14

Kimi K2.7 Code

Runs on Data centre · Hugging Face ↗

vLLMSGLang

Coding index 60.8 (Artificial Analysis)

Moonshot's ~1-trillion-parameter MoE, tuned specifically for coding — and unlike Kimi K3 the weights ARE openly released. One of the strongest open coders.

Sizes on disk

Real GGUF file sizes = weight VRAM.

QuantizationSize
Q2_K339.5 GB
Q3_K_M463.6 GB
IQ4_XS495.1 GB
Q4583.7 GB
Q8594.5 GB

Config tips

~495 GB at Q4 — cluster-scale, tensor / pipeline parallel across many GPUs. Open but not local for most; the hosted per-token price is the realistic route.