Back to search results
Qwen3.5-4B · Q4_K_M
Apple M2 Prollama.cpp v0.49L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-4B (Q4_K_M) on Apple M2 Pro with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
94 tok/s
Decode
Decode speed
499.11 tok/s
Prefill
Prefill speed
1.42 s
TTFT
Time to first token
8.99 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_M Decode 118.95 · Prefill 519.35 ·
- llama.cpp · Q4_K_M (current)Decode 94 · Prefill 499.11 ·
- Ollama · Q6_K Decode 70.16 · Prefill 240.68 ·
Core figures
- Decode (avg)
- 94 tok/s
- Prefill (avg)
- 499.11 tok/s
- TTFT (avg)
- 1.42 s
- VRAM (avg)
- 8.99 GB
- MTP acceptance rate
- —
- TTFB
- 0.75 GB
- Power draw
- 65.7 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-4B
- Quantization
- Q4_K_M
- Framework
- llama.cpp
- Version
- v0.49
- Context length
- 16384 tokens
- Batch size
- 512
- GPU layers
- 33
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- Apple M2 Pro
- Measured VRAM (avg)
- 8.99 GB
- OS
- macOS 15.0
- Driver
- Apple M2 Pro
- CUDA
- N/A
- Power draw
- 65.7 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.49 · macOS 15.0 · CUDA N/A · 16384 ctx
94 tok/s
Decode
499.11 tok/s
Prefill
1.42 s
TTFT
8.99 GB
VRAM
—
MTP
65.7 W
Power draw
L2 batch seed 2026-10-16: llama.cpp on hw_id=102 running model_id=48 at Q4_K_M