Back to search results
Qwen2.5-Omni-3B · IQ3_S
2x AMD Strix Halollama.cpp v0.45L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen2.5-Omni-3B (IQ3_S) on 2x AMD Strix Halo with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
239.07 tok/s
Decode
Decode speed
794.3 tok/s
Prefill
Prefill speed
0.91 s
TTFT
Time to first token
6.26 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 (current)Decode 239.07 · Prefill 794.3 ·
- vLLM · Q5_K_M Decode 183.47 · Prefill 864.96 ·
- vLLM · Q6_K Decode 171.22 · Prefill 756.33 ·
Core figures
- Decode (avg)
- 239.07 tok/s
- Prefill (avg)
- 794.3 tok/s
- TTFT (avg)
- 0.91 s
- VRAM (avg)
- 6.26 GB
- MTP acceptance rate
- —
- TTFB
- 1.91 GB
- Power draw
- 86.3 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen2.5-Omni-3B
- Quantization
- IQ3_S
- Framework
- llama.cpp
- Version
- v0.45
- Context length
- 8192 tokens
- Batch size
- 256
- GPU layers
- 28
- Flash Attention
- On
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)
- 6.26 GB
- OS
- Windows 11 23H2
- Driver
- amdgpu-pro 24.20
- CUDA
- N/A
- Power draw
- 86.3 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.45 · Windows 11 23H2 · CUDA N/A · 8192 ctx
239.07 tok/s
Decode
794.3 tok/s
Prefill
0.91 s
TTFT
6.26 GB
VRAM
—
MTP
86.3 W
Power draw
L2 batch seed 2026-10-16: llama.cpp on hw_id=88 running model_id=60 at IQ3_S