Back to search results
Qwen3-32B · Q4_K_M
NVIDIA RTX 409024GBOllama 0.5.7L1 Reproduced
This page aggregates 1 real-world runs of Qwen3-32B (Q4_K_M) on NVIDIA RTX 4090 with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.
17.4 tok/s
Decode
Decode speed
198.2 tok/s
Prefill
Prefill speed
1.31 s
TTFT
Time to first token
19.8 GB
VRAM
VRAM usage
L1 Reproduced
1
Measured runs
1
Independent sources
V2EX
Source platforms
23 days ago
Last verified
Performance
- llama.cpp · Q4_K_M Decode 18.4 · Prefill 216.93 ·
- Ollama · Q4_K_M (current)Decode 17.4 · Prefill 198.2 ·
Core figures
- Decode (avg)
- 17.4 tok/s
- Prefill (avg)
- 198.2 tok/s
- TTFT (avg)
- 1.31 s
- VRAM (avg)
- 19.8 GB
- MTP acceptance rate
- —
- TTFB
- — GB
- Power draw
- 339 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3-32B
- Quantization
- Q4_K_M
- Framework
- Ollama
- Version
- 0.5.7
- Context length
- 8192 tokens
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 4090
- Nominal VRAM
- 24 GB
- Measured VRAM (avg)
- 19.8 GB
- OS
- Ubuntu 24.04
- Driver
- 550.107.02
- CUDA
- 12.4
- Power draw
- 339 W
Sources & evidence
1 measured records in total, each traceable to its original source
0.5.7 · Ubuntu 24.04 · CUDA 12.4 · 8192 ctx
17.4 tok/s
Decode
198.2 tok/s
Prefill
1.31 s
TTFT
19.8 GB
VRAM
—
MTP
339 W
Power draw
Ollama 默认参数下比 llama.cpp 直跑略低。