TEST #4138 RESULTS
09/09/2026 - 4:33 PM
ACCELERATOR
MODEL
3.6
tokens/s
generation
146.30
sec
time to first token
10
tokens/s
prompt
6
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Core Processor (Haswell, no TSX) (haswell)
RAM
7.6GB
OS
Linux
Kernel Release
7.0.6-2-pve
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC PMX 7.0.6-2 (2026-05-29T11:08Z)
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
10
tokens/s
3.7
tokens/s
107.02
sec
pp4096+tg256
9
tokens/s
2.9
tokens/s
478.07
sec
pp2048+tg256
9
tokens/s
3.4
tokens/s
224.35
sec
pp2048+tg768
9
tokens/s
3.4
tokens/s
221.86
sec
pp1024+tg1024
10
tokens/s
3.6
tokens/s
105.44
sec
pp1280+tg3072
10
tokens/s
3.1
tokens/s
130.60
sec
pp384+tg1152
10
tokens/s
3.7
tokens/s
40.65
sec
pp64+tg1024
10
tokens/s
4.2
tokens/s
6.98
sec
pp16+tg1536
11
tokens/s
4.0
tokens/s
1.74
sec