TEST #3680 RESULTS
06/03/2026 - 3:32 PM
ACCELERATOR
MODEL
36.5
tokens/s
generation
658
ms
time to first token
2106
tokens/s
prompt
489
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
RAM
121.6GB
OS
Linux
Kernel Release
6.17.0-1018-nvidia
Architecture
aarch64
Version
Cosmopolitan 3.9.7 MODE=aarch64; #18-Ubuntu SMP PREEMPT_DYNAMIC Tue May 5 21:28:33 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2564
tokens/s
37.5
tokens/s
427
ms
pp4096+tg256
1761
tokens/s
33.1
tokens/s
2.36
sec
pp2048+tg256
2227
tokens/s
36.0
tokens/s
948
ms
pp2048+tg768
2232
tokens/s
35.6
tokens/s
946
ms
pp1024+tg1024
2571
tokens/s
36.9
tokens/s
426
ms
pp1280+tg3072
2497
tokens/s
34.9
tokens/s
540
ms
pp384+tg1152
3072
tokens/s
37.8
tokens/s
152
ms
pp64+tg1024
1548
tokens/s
38.4
tokens/s
68
ms
pp16+tg1536
479
tokens/s
38.1
tokens/s
60
ms