TEST #3022 RESULTS
03/07/2026 - 10:20 PM
ACCELERATOR
MODEL
59.0
tokens/s
generation
671
ms
time to first token
1963
tokens/s
prompt
557
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2699A v4 @ 2.40GHz (broadwell)
RAM
503.7GB
OS
Linux
Kernel Release
6.18.9-arch1-2.1
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Wed, 11 Feb 2026 13:53:26 +0000
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2291
tokens/s
62.2
tokens/s
463
ms
pp4096+tg256
1837
tokens/s
54.6
tokens/s
2.25
sec
pp2048+tg256
2120
tokens/s
59.0
tokens/s
983
ms
pp2048+tg768
2097
tokens/s
57.9
tokens/s
994
ms
pp1024+tg1024
2259
tokens/s
59.8
tokens/s
470
ms
pp1280+tg3072
2159
tokens/s
55.9
tokens/s
609
ms
pp384+tg1152
2388
tokens/s
60.4
tokens/s
177
ms
pp64+tg1024
2040
tokens/s
61.1
tokens/s
47
ms
pp16+tg1536
478
tokens/s
60.5
tokens/s
49
ms