Qwen3.8-27B · Q5_K_S
This page aggregates 1 real-world runs of Qwen3.8-27B (Q5_K_S) on Apple M4 Max with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
122.2 tok/s
Decode
Decode speed
481.39 tok/s
Prefill
Prefill speed
1.01 s
TTFT
Time to first token
53.13 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- vLLM · Q5_K_S (current)Decode 122.2 · Prefill 481.39 ·
- Strata · IQ2_M Decode 265.43 · Prefill 1342.3 ·
- Strata · IQ3_S Decode 228.14 · Prefill 1118.4 ·
- llama.cpp · Q4_K_M Decode 26 · Prefill 195 ·
Core figures
- Decode (avg)
- 122.2 tok/s
- Prefill (avg)
- 481.39 tok/s
- TTFT (avg)
- 1.01 s
- VRAM (avg)
- 53.13 GB
- MTP acceptance rate
- —
- TTFB
- 0.98 GB
- Power draw
- 45.7 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.8-27B
- Quantization
- Q5_K_S
- Framework
- vLLM
- Version
- v0.26
- Context length
- 16384 tokens
- Batch size
- 32
- GPU layers
- 999
- Flash Attention
- On
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)
- 53.13 GB
- OS
- macOS 15.0
- Driver
- Apple M4 Max
- CUDA
- N/A
- Power draw
- 45.7 W
Get started
Original model resources
Base model resources may not match this quantization. Use the verified deployment weights above when available.
Sources & evidence
1 measured records in total, each traceable to its original source
v0.26 · macOS 15.0 · CUDA N/A · 16384 ctx
122.2 tok/s
Decode
481.39 tok/s
Prefill
1.01 s
TTFT
53.13 GB
VRAM
—
MTP
45.7 W
Power draw
L2 batch seed 2026-10-10: vLLM on hw_id=26 running model_id=5 at Q5_K_S