community pool / model answer

Laguna-S-2.1

chat · poolside/Laguna-S-2.1

41.9 tok/s · model median · 47 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
Multi-GPU 72GB ×3	4-bit	60.3 tok/s	measured	327	65.1 GB	278,528	llama.cpp
Radeon AI Pro R9700 32GB ×3	4-bit	48.3 tok/s	measured	780	65.5 GB	131,072	llama.cpp
RTX PRO 6000 Blackwell 96GB ×4	16-bit	126.6 tok/s	reported	50	- GB	2,048	vllm
Ryzen AI Max 395 128GB	4-bit	40.7 tok/s	measured	2,811	- GB	262,144	llama.cpp
Ryzen AI Max 395 64GB	2-bit	43.7 tok/s	measured	2,640	35.9 GB	10,240	llama.cpp
Ryzen AI Max 395 64GB	3-bit	32.8 tok/s	measured	2,823	52 GB	10,240	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.