Each row is a run configuration (model + hardware + framework + quantization); repeated runs of the same setup are merged — raw counts appear in each result’s source line.
Intel Arc B580 · 12GB · llama.cpp
Other
Today
— tok/s
Decode
— tok/s
Prefill
— s
TTFT
6 GB
VRAM
Sources: Other · 1 runs · 1 independent sources
NVIDIA RTX 5070 · 12GB · Strata
Other
Today
137.1 tok/s
Decode
1,028.3 tok/s
Prefill
0.1 s
TTFT
5.4 GB
VRAM
Sources: Other · 1 runs · 1 independent sources
NVIDIA RTX 3060 12GB · 12GB · llama.cpp
Other
Today
51.1 tok/s
Decode
383.3 tok/s
Prefill
0.26 s
TTFT
6 GB
VRAM
Sources: Other · 1 runs · 1 independent sources
3 results