community pool / model answer

Ornith-1.0-35B-GGUF

chat · ornith-ai/Ornith-1.0-35B-GGUF

60.6 tok/s · model median · 30 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
Arc Pro B60 24GB	4-bit	62.7 tok/s	reported	532	- GB	4,096	llama.cpp
Arc Pro B60 24GB ×2	4-bit	62.6 tok/s	reported	532	- GB	4,096	llama.cpp
RTX 3060 12GB ×2	4-bit	85.9 tok/s	measured	3,079	21.1 GB	65,536	llama.cpp
RTX 3090 24GB	4-bit	124.1 tok/s	reported	94	- GB	262,144	llama.cpp
RTX 3090 24GB ×2	4-bit	136.5 tok/s	reported	86	- GB	262,144	llama.cpp
RTX 4070 12GB	4-bit	43.2 tok/s	measured	393	- GB	2,048	vllm
RTX 5090 32GB	4-bit	252.6 tok/s	reported	1	- GB	2,048	ollama
RX 5700 4GB	4-bit	27.4 tok/s	reported	122	- GB	2,048	llama.cpp
RX 5700 8GB	4-bit	27.4 tok/s	reported	127	- GB	2,048	llama.cpp
Ryzen AI Max 395 128GB	4-bit	43 tok/s	reported	26,649	20.9 GB	65,536	llama.cpp
Ryzen AI Max 395 64GB	4-bit	60.6 tok/s	measured	1,583	21.5 GB	8,192	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.