DiggTop

RTX 3090 vs Tesla V100 x4

Best verified output tok/s per model (single-stream generation). Hover a value for the quant.

RTX 309012 verified runs · best 709.5 tok/s
Tesla V100 x41 verified run · best 391.0 tok/s
ModelRTX 3090Tesla V100 x4
LFM2.5-1.2B-Instruct-GGUF709.5 Q4_K_M
Qwen3.5-0.8B-MTP-GGUF476.1 Q6_K_XL
Qwen3.8-27B152.4 EXL3 4.00bpw391.0 NVFP4
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF353.7 NVFP4
LFM2.5-2.6B-GGUF305.6 Q4_K_M
Qwen3.5-4B-MTP-GGUF240.4 Q6_K_XL
Qwen3.8-27B-EXL3-3.5bpw164.6 EXL3-3.5bpw
Qwen3.5-9B-MTP-GGUF164.3 Q6_K_XL
Qwen/Qwen3.8-27B136.8 AWQ-INT4 W4A16
Qwen3.8-27B-GGUF81.9 Q4_K_XL
Ornstein3.6-27B-MTP-NSC-ACE-SABER-GGUF32.9 Q4_K_M

Head-to-head (models tested on both): RTX 3090 0 · Tesla V100 x4 1 · tied 0 · data as of Sep 16, 2026. Full rules: methodology.

Related reading: RTX 5090 vs 3090 · quantization · Value index · Full board