community pool / model answer

LFM2.5-1.2B-Instruct-GGUF

chat · LiquidAI/LFM2.5-1.2B-Instruct-GGUF

120.3 tok/s · model median · 31 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
M5 Pro 24GB	4-bit	267.8 tok/s	reported	90	- GB	4,224	llama.cpp
RTX 3090 24GB	4-bit	682.2 tok/s	measured	25	2.4 GB	24,576	llama.cpp
RTX 3090 24GB	6-bit	564.4 tok/s	measured	28	2.6 GB	24,576	llama.cpp
RX 570 4GB	4-bit	118 tok/s	measured	26	- GB	2,048	llama.cpp
RX 570 4GB ×2	4-bit	120.6 tok/s	measured	22	- GB	2,048	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.