Back to search results
Qwen3.5-27B · Q4_K_S
Apple M5 Max128GBllama.cpp v0.44L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-27B (Q4_K_S) on Apple M5 Max with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
146.98 tok/s
Decode
Decode speed
610.13 tok/s
Prefill
Prefill speed
0.12 s
TTFT
Time to first token
46.33 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 · IQ3_XS Decode 171.72 · Prefill 594.08 ·
- llama.cpp · Q4_K_S (current)Decode 146.98 · Prefill 610.13 ·
- Ollama · Q5_K_M Decode 119.42 · Prefill 542.35 ·
- Ollama · Q6_K Decode 100.18 · Prefill 374.17 ·
Core figures
- Decode (avg)
- 146.98 tok/s
- Prefill (avg)
- 610.13 tok/s
- TTFT (avg)
- 0.12 s
- VRAM (avg)
- 46.33 GB
- MTP acceptance rate
- —
- TTFB
- 1.18 GB
- Power draw
- 183.2 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-27B
- Quantization
- Q4_K_S
- Framework
- llama.cpp
- Version
- v0.44
- Context length
- 4096 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
- Apple M5 Max
- Nominal VRAM
- 128 GB
- Measured VRAM (avg)
- 46.33 GB
- OS
- Ubuntu 22.04
- Driver
- 570.86
- CUDA
- 12.6
- Power draw
- 183.2 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.44 · Ubuntu 22.04 · CUDA 12.6 · 4096 ctx
146.98 tok/s
Decode
610.13 tok/s
Prefill
0.12 s
TTFT
46.33 GB
VRAM
—
MTP
183.2 W
Power draw
L2 batch seed 2026-10-16: llama.cpp on hw_id=28 running model_id=50 at Q4_K_S