Gemma 4 26B A4B IT adapted for high-quality local analyst workflows with about 4B active parameters per token.

Direct base-model GGUF (Q4_K_M) of google/gemma-4-26B-A4B-it, published for TribeBlend's local Data Chat runtime. TribeBlend grounds answers with Knowledge Graph context at prompt time and a model-aware agent harness, so the base instruction/reasoning model ships as-is (no fine-tuning).

Designed for TribeBlend Data Chat, loaded via llama-cpp-2.

Inherits the upstream base-model license (other); verify upstream terms before redistribution.