Back to search results
Qwen2.5-0.5B-Instruct · Q3_K_M
AMD Ryzen AI Max+ 395128GBllama.cpp v0.31L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen2.5-0.5B-Instruct (Q3_K_M) on AMD Ryzen AI Max+ 395 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
339.19 tok/s
Decode
Decode speed
1,541.81 tok/s
Prefill
Prefill speed
0.96 s
TTFT
Time to first token
2.81 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 · Q3_K_M (current)Decode 339.19 · Prefill 1541.81 ·
- llama.cpp · Q3_K_S Decode 367.02 · Prefill 1342.6 ·
- Strata · IQ2_M Decode 347.23 · Prefill 1400.32 ·
- Ollama · Q3_K_M Decode 320.69 · Prefill 1372.43 ·
- Strata · IQ1_S Decode 308.91 · Prefill 1621.83 ·
- Strata · Q3_K_M Decode 316.37 · Prefill 1349.83 ·
Core figures
- Decode (avg)
- 339.19 tok/s
- Prefill (avg)
- 1,541.81 tok/s
- TTFT (avg)
- 0.96 s
- VRAM (avg)
- 2.81 GB
- MTP acceptance rate
- —
- TTFB
- 2.71 GB
- Power draw
- 94.6 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen2.5-0.5B-Instruct
- Quantization
- Q3_K_M
- Framework
- llama.cpp
- Version
- v0.31
- Context length
- 16384 tokens
- Batch size
- 256
- GPU layers
- 20
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- AMD Ryzen AI Max+ 395
- Nominal VRAM
- 128 GB
- Measured VRAM (avg)
- 2.81 GB
- OS
- Ubuntu 22.04
- Driver
- amdgpu-pro 24.20
- CUDA
- N/A
- Power draw
- 94.6 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.31 · Ubuntu 22.04 · CUDA N/A · 16384 ctx
339.19 tok/s
Decode
1,541.81 tok/s
Prefill
0.96 s
TTFT
2.81 GB
VRAM
—
MTP
94.6 W
Power draw
L2 batch seed 2026-10-16: llama.cpp on hw_id=24 running model_id=59 at Q3_K_M