Back to search results
Qwen 4 72B · Q6_K
NVIDIA RTX 509032GBvLLM 0.7.3L0 Self-reported
This page aggregates 1 real-world runs of Qwen 4 72B (Q6_K) on NVIDIA RTX 5090 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
24.1 tok/s
Decode
Decode speed
180.8 tok/s
Prefill
Prefill speed
0.55 s
TTFT
Time to first token
32 GB
VRAM
VRAM usage
L0 Self-reported
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- vLLM · IQ3_S Decode 52.2 · Prefill 391.5 ·
- vLLM · IQ4_XS Decode 39.2 · Prefill 294 ·
- vLLM · Q6_K (current)Decode 24.1 · Prefill 180.8 ·
- vLLM · Q5_K_M Decode 28.5 · Prefill 213.8 ·
- llama.cpp · IQ3_S Decode 47.9 · Prefill 359.3 ·
- llama.cpp · IQ4_XS Decode 35.9 · Prefill 269.3 ·
- llama.cpp · Q6_K Decode 22.1 · Prefill 165.8 ·
- llama.cpp · Q5_K_M Decode 26.1 · Prefill 195.8 ·
- vLLM · Q4_K_M Decode 34.9 · Prefill 261.8 ·
Core figures
- Decode (avg)
- 24.1 tok/s
- Prefill (avg)
- 180.8 tok/s
- TTFT (avg)
- 0.55 s
- VRAM (avg)
- 32 GB
- MTP acceptance rate
- —
- TTFB
- — GB
- Power draw
- — W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen 4 72B
- Quantization
- Q6_K
- Framework
- vLLM
- Version
- 0.7.3
- Context length
- 4096 tokens
- Batch size
- 512
- GPU layers
- 99
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 5090
- Nominal VRAM
- 32 GB
- Measured VRAM (avg)
- 32 GB
- OS
- Linux
- Driver
- 570.65
- CUDA
- 12.6
Sources & evidence
1 measured records in total, each traceable to its original source
0.7.3 · Linux · CUDA 12.6 · 4096 ctx
24.1 tok/s
Decode
180.8 tok/s
Prefill
0.55 s
TTFT
32 GB
VRAM
—
MTP
— W
Power draw
L0 理论估算（add-1000-records-batch3）