Back to search results
Qwen3.5-27B · Q3_K_M
Apple M4 Max128GBvLLM v0.54L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-27B (Q3_K_M) on Apple M4 Max with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
166.04 tok/s
Decode
Decode speed
735.49 tok/s
Prefill
Prefill speed
2.33 s
TTFT
Time to first token
40.66 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · IQ3_S Decode 165.6 · Prefill 722.22 ·
- vLLM · Q3_K_M (current)Decode 166.04 · Prefill 735.49 ·
- Ollama · Q3_K_S Decode 165.47 · Prefill 649.98 ·
- Ollama · Q5_K_S Decode 111.59 · Prefill 447.41 ·
Core figures
- Decode (avg)
- 166.04 tok/s
- Prefill (avg)
- 735.49 tok/s
- TTFT (avg)
- 2.33 s
- VRAM (avg)
- 40.66 GB
- MTP acceptance rate
- —
- TTFB
- 1.33 GB
- Power draw
- 68.7 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-27B
- Quantization
- Q3_K_M
- Framework
- vLLM
- Version
- v0.54
- Context length
- 4096 tokens
- Batch size
- 16
- GPU layers
- 999
- Flash Attention
- Off
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- Apple M4 Max
- Nominal VRAM
- 128 GB
- Measured VRAM (avg)
- 40.66 GB
- OS
- macOS 15.0
- Driver
- Apple M4 Max
- CUDA
- N/A
- Power draw
- 68.7 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.54 · macOS 15.0 · CUDA N/A · 4096 ctx
166.04 tok/s
Decode
735.49 tok/s
Prefill
2.33 s
TTFT
40.66 GB
VRAM
—
MTP
68.7 W
Power draw
L2 batch seed 2026-10-10: vLLM on hw_id=26 running model_id=50 at Q3_K_M