TEST #29 RESULTS
03/27/2025 - 11:54 PM
ACCELERATOR
MODEL
38.8
tokens/s
generation
1.12
sec
time to first token
1116
tokens/s
prompt
338
LocalScore
HOW YOU STACK UP
Explore All ResultsMistral Small 24B Instruct 2501 - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7H12 64-Core Processor (znver2)
RAM
503.5GB
OS
Linux
Kernel Release
6.8.0-52-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #53~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Jan 15 19:18:46 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1289
tokens/s
39.9
tokens/s
819
ms
pp4096+tg256
1167
tokens/s
36.3
tokens/s
3.54
sec
pp2048+tg256
1238
tokens/s
38.4
tokens/s
1.68
sec
pp2048+tg768
1243
tokens/s
38.2
tokens/s
1.67
sec
pp1024+tg1024
1277
tokens/s
39.1
tokens/s
827
ms
pp1280+tg3072
1253
tokens/s
37.5
tokens/s
1.05
sec
pp384+tg1152
1238
tokens/s
39.5
tokens/s
335
ms
pp64+tg1024
1001
tokens/s
40.0
tokens/s
88
ms
pp16+tg1536
339
tokens/s
39.8
tokens/s
71
ms