TEST #4100 RESULTS
08/28/2026 - 7:44 PM
ACCELERATOR
MODEL
29.9
tokens/s
generation
6.73
sec
time to first token
208
tokens/s
prompt
97
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Genuine Intel CPU 0000%@ (sapphirerapids)
RAM
125.5GB
OS
Linux
Kernel Release
6.12.105+deb13-amd64
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Debian 6.12.105-1 (2026-08-24)
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
208
tokens/s
32.3
tokens/s
4.95
sec
pp4096+tg256
181
tokens/s
30.3
tokens/s
22.63
sec
pp2048+tg256
207
tokens/s
33.2
tokens/s
9.91
sec
pp2048+tg768
206
tokens/s
26.8
tokens/s
9.99
sec
pp1024+tg1024
215
tokens/s
20.9
tokens/s
4.80
sec
pp1280+tg3072
210
tokens/s
31.2
tokens/s
6.13
sec
pp384+tg1152
220
tokens/s
28.6
tokens/s
1.78
sec
pp64+tg1024
245
tokens/s
33.0
tokens/s
292
ms
pp16+tg1536
179
tokens/s
32.5
tokens/s
120
ms