community pool / model answer

gemma-4-26B-A4B-it-qat-GGUF

chat · unsloth/gemma-4-26B-A4B-it-qat-GGUF

95.3 tok/s · model median · 42 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
M1 Max 64GB	4-bit	76.4 tok/s	reported	50	- GB	2,048	ollama
M5 Pro 24GB	4-bit	121.6 tok/s	reported	147	- GB	2,048	llama.cpp
RTX 3050 8GB	4-bit	38 tok/s	reported	952	- GB	28,000	llama.cpp
RTX 3060 12GB ×2	4-bit	87.8 tok/s	measured	2,379	16 GB	131,072	llama.cpp
RTX 3090 24GB	4-bit	138.3 tok/s	measured	3,433	0 GB	131,072	llama.cpp

Context tested

The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.