TEST #2590 RESULTS
02/10/2026 - 2:00 PM
ACCELERATOR
MODEL
63.4
tokens/s
generation
712
ms
time to first token
1830
tokens/s
prompt
546
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 3950X 16-Core Processor (znver2)
RAM
94.2GB
OS
Linux
Kernel Release
6.8.0-94-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #96-Ubuntu SMP PREEMPT_DYNAMIC Fri Jan 9 20:36:55 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2153
tokens/s
64.9
tokens/s
491
ms
pp4096+tg256
1717
tokens/s
57.3
tokens/s
2.40
sec
pp2048+tg256
1985
tokens/s
62.1
tokens/s
1.05
sec
pp2048+tg768
1978
tokens/s
61.3
tokens/s
1.05
sec
pp1024+tg1024
2139
tokens/s
63.5
tokens/s
494
ms
pp1280+tg3072
2060
tokens/s
59.9
tokens/s
637
ms
pp384+tg1152
2222
tokens/s
66.7
tokens/s
188
ms
pp64+tg1024
1691
tokens/s
67.8
tokens/s
52
ms
pp16+tg1536
527
tokens/s
67.2
tokens/s
45
ms