Quantization bit depths are assigned per-tensor using sensitivity probing rather than a uniform scheme. Each attention tensor receives bits proportional to how much its output diverges under quantization noise, measured across 8 random probes. Expert tensors (384 routed experts × 60 MoE layers) are quantized uniformly at Q2_K.

Architecture: DeepSeek-V3 decoder (kimi_k2), 61 layers, 384 routed experts / 8 active per token, MLA attention, hidden_size=7168, 1.03T total parameters.

Usage (llama.cpp)

bash

llama-cli \
  -m kimi-k2.6-ram-344gb.gguf \
  -c 8192 \
  --temp 0.6 \
  -p "You are a helpful assistant."

Notes

No importance matrix (imatrix) was used — the 1.1 TB Q8_0 intermediate required for imatrix generation exceeds available RAM for inference on this machine. Bit allocation from sensitivity probing provides the primary quality signal.
The original model uses Neural Magic compressed-tensors INT4 for expert weights. These are dequantized and re-quantized to GGUF format during conversion.

baa-ai/Kimi-K2.6-RAM-344GB-GGUF

Kimi-K2.6 RAM GGUF — 344 GB (Q2_K experts)

File

Method

Usage (llama.cpp)

Notes

baa-ai/
Kimi-K2.6-RAM-344GB-GGUF