Back to search results
Qwen2.5-Omni-3B · Q5_K_M
2x AMD Strix HalovLLM v0.37L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen2.5-Omni-3B (Q5_K_M) on 2x AMD Strix Halo with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
183.47 tok/s
Decode
Decode speed
864.96 tok/s
Prefill
Prefill speed
0.57 s
TTFT
Time to first token
7.96 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
2 days ago
Last verified
Performance
- llama.cpp · IQ2_XS Decode 273.09 · Prefill 1000.97 ·
- llama.cpp · IQ3_S Decode 239.07 · Prefill 794.3 ·
- vLLM · Q5_K_M (current)Decode 183.47 · Prefill 864.96 ·
- vLLM · Q6_K Decode 171.22 · Prefill 756.33 ·
Core figures
- Decode (avg)
- 183.47 tok/s
- Prefill (avg)
- 864.96 tok/s
- TTFT (avg)
- 0.57 s
- VRAM (avg)
- 7.96 GB
- MTP acceptance rate
- —
- TTFB
- 1.47 GB
- Power draw
- 60 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen2.5-Omni-3B
- Quantization
- Q5_K_M
- Framework
- vLLM
- Version
- v0.37
- Context length
- 16384 tokens
- Batch size
- 32
- GPU layers
- 999
- Flash Attention
- Off
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- 2x AMD Strix Halo
- Measured VRAM (avg)
- 7.96 GB
- OS
- Ubuntu 24.04
- Driver
- 560.94
- CUDA
- N/A
- Power draw
- 60 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.37 · Ubuntu 24.04 · CUDA N/A · 16384 ctx
183.47 tok/s
Decode
864.96 tok/s
Prefill
0.57 s
TTFT
7.96 GB
VRAM
—
MTP
60 W
Power draw
L2 batch seed 2026-10-16: vLLM on hw_id=88 running model_id=60 at Q5_K_M