TEST #3609 RESULTS
05/25/2026 - 2:44 PM
ACCELERATOR
MODEL
57.4
tokens/s
generation
887
ms
time to first token
1428
tokens/s
prompt
452
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2660 v3 @ 2.60GHz (haswell)
RAM
247.5GB
OS
Linux
Kernel Release
7.0.0-15-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #15-Ubuntu SMP PREEMPT_DYNAMIC Wed Apr 22 16:06:43 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1726
tokens/s
62.3
tokens/s
610
ms
pp4096+tg256
1418
tokens/s
49.1
tokens/s
2.91
sec
pp2048+tg256
1578
tokens/s
56.3
tokens/s
1.32
sec
pp2048+tg768
1571
tokens/s
55.2
tokens/s
1.32
sec
pp1024+tg1024
1648
tokens/s
58.9
tokens/s
638
ms
pp1280+tg3072
1599
tokens/s
52.7
tokens/s
818
ms
pp384+tg1152
1615
tokens/s
60.1
tokens/s
254
ms
pp64+tg1024
1273
tokens/s
61.2
tokens/s
66
ms
pp16+tg1536
424
tokens/s
60.8
tokens/s
53
ms