Back to search results
Qwen3.5-4B · Q3_K_M
Intel Arc A770 16GB16GBllama.cpp v0.51L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-4B (Q3_K_M) on Intel Arc A770 16GB with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
75.53 tok/s
Decode
Decode speed
384.12 tok/s
Prefill
Prefill speed
0.62 s
TTFT
Time to first token
7.81 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
5 days ago
Last verified
Performance
- vLLM · Q3_K_M Decode 77.24 · Prefill 250.09 ·
- vLLM · Q4_K_M Decode 74.79 · Prefill 395.91 ·
- Ollama · Q2_K Decode 90.54 · Prefill 350.46 ·
- Ollama · Q5_K_S Decode 56.42 · Prefill 256.71 ·
- llama.cpp · Q3_K_M (current)Decode 75.53 · Prefill 384.12 ·
- llama.cpp · Q4_K_M Decode 65.4 · Prefill 252.99 ·
- vLLM · Q3_K_L Decode 82.73 · Prefill 332.62 ·
- vLLM · Q8_0 Decode 45.45 · Prefill 151.34 ·
Core figures
- Decode (avg)
- 75.53 tok/s
- Prefill (avg)
- 384.12 tok/s
- TTFT (avg)
- 0.62 s
- VRAM (avg)
- 7.81 GB
- MTP acceptance rate
- —
- TTFB
- 0.6 GB
- Power draw
- 227.3 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-4B
- Quantization
- Q3_K_M
- Framework
- llama.cpp
- Version
- v0.51
- Context length
- 4096 tokens
- Batch size
- 512
- GPU layers
- 33
- Flash Attention
- Off
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- Intel Arc A770 16GB
- Nominal VRAM
- 16 GB
- Measured VRAM (avg)
- 7.81 GB
- OS
- Ubuntu 24.04
- Driver
- Intel Arc 31.0.101.5186
- CUDA
- 12.6
- Power draw
- 227.3 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.51 · Ubuntu 24.04 · CUDA 12.6 · 4096 ctx
75.53 tok/s
Decode
384.12 tok/s
Prefill
0.62 s
TTFT
7.81 GB
VRAM
—
MTP
227.3 W
Power draw
L2 batch seed 2026-10-13: llama.cpp on hw_id=97 running model_id=48 at Q3_K_M