Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

LLM Inference Performance on LLaMA-3 8B

56.03TTFT (ms)

MXFP4

-141.21081,190.16462,521.543,852.9154Mar 9, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
56.032.366.942.5125.480.72
2026.03
62.312.127.032.4839.280.47
2026.03
132.38117.44118.441
2026.03
542.582.3117.41.654.491.14
2026.03
587.172.1317.481.657.081.08
2026.03
1,113.062.2629.351.36100.711.13
2026.03
1,206.692.0829.441.35103.271.11
2026.03
1,252.24127.9161.931
2026.03
2,257.542.2153.251.2193.71.13
2026.03
2,453.292.0353.341.2202.131.08
2026.03
2,515.08139.851114.241
2026.03
4,987.05163.761218.011