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

LLM Inference on Qwen3-8B (2k prompts Balanced workload)

61,106Throughput (tok/s)

v1

6,939.6821,002.0935,064.549,126.91May 30, 2026
Updated 1d ago

Evaluation Results

MethodLinks
2026.05
61,10646414.7
2026.05
60,98841314.9
2026.05
48,8951,12817.3
2026.05
47,4336818.9
2026.05
46,2615,7539.2
2026.05
43,72915210.9
2026.05
43,64815810.9
2026.05
42,01639510.9
2026.05
31,3412497.5
2026.05
29,3881788.2
2026.05
28,686748.6
2026.05
28,646758.6
2026.05
26,45559617.6
2026.05
25,0804,01329.9
2026.05
24,03769738.6
2026.05
23,91469738.7
2026.05
21,96323311
2026.05
20,3971,67643.4
2026.05
19,83184422.9
2026.05
17,55150526.9
2026.05
16,91323528.6
2026.05
16,87022928.7
2026.05
12,62931619.2
2026.05
11,64022921.1
2026.05
11,40266442.1
2026.05
11,06311622.5
2026.05
11,01312122.6
2026.05
9,02327227.3