TEST #4178 RESULTS
09/17/2026 - 8:02 AM
ACCELERATOR
MODEL
48.9
tokens/s
generation
602
ms
time to first token
2305
tokens/s
prompt
572
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 5900X 12-Core Processor (znver3)
RAM
31.3GB
OS
Linux
Kernel Release
6.8.0-139-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #139-Ubuntu SMP PREEMPT_DYNAMIC Sat Aug 1 03:52:05 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2739
tokens/s
50.0
tokens/s
394
ms
pp4096+tg256
1936
tokens/s
45.6
tokens/s
2.14
sec
pp2048+tg256
2404
tokens/s
48.3
tokens/s
873
ms
pp2048+tg768
2403
tokens/s
47.9
tokens/s
873
ms
pp1024+tg1024
2726
tokens/s
49.3
tokens/s
396
ms
pp1280+tg3072
2660
tokens/s
47.5
tokens/s
501
ms
pp384+tg1152
3113
tokens/s
50.1
tokens/s
143
ms
pp64+tg1024
2234
tokens/s
50.7
tokens/s
48
ms
pp16+tg1536
535
tokens/s
50.4
tokens/s
49
ms