community pool / model answer

Ornith-1.0-9B

chat · ornith-ai/Ornith-1.0-9B

25.7 tok/s · model median · 44 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
Arc Pro B70 32GB	6-bit	50.3 tok/s	reported	58	- GB	262,144	llama.cpp
M3 Ultra 512GB	4-bit	77.5 tok/s	reported	104	- GB	8,192	llama.cpp
Radeon AI Pro R9700 32GB ×3	16-bit	25.7 tok/s	measured	394	25.8 GB	787	llama.cpp
RX 570 4GB	4-bit	5.3 tok/s	measured	411	- GB	2,048	llama.cpp
RX 570 4GB ×2	4-bit	23.8 tok/s	measured	157	- GB	2,048	llama.cpp
RX 6700 XT 12GB	8-bit	37.2 tok/s	measured	-	- GB	4,736	llama.cpp
RX 7700 XT 12GB	5-bit	30.6 tok/s	reported	87	- GB	2,048	vllm

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.