Back to search results
Qwen3.5-4B · FP16
Apple M1 MaxvLLM v0.23L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-4B (FP16) on Apple M1 Max with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
57.12 tok/s
Decode
Decode speed
260.15 tok/s
Prefill
Prefill speed
1.28 s
TTFT
Time to first token
17.64 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
2 days ago
Last verified
Performance
- vLLM · FP16 (current)Decode 57.12 · Prefill 260.15 ·
- vLLM · Q3_K_S Decode 135.63 · Prefill 662.69 ·
- Strata · IQ4_XS Decode 158.99 · Prefill 630.94 ·
Core figures
- Decode (avg)
- 57.12 tok/s
- Prefill (avg)
- 260.15 tok/s
- TTFT (avg)
- 1.28 s
- VRAM (avg)
- 17.64 GB
- MTP acceptance rate
- —
- TTFB
- 0.46 GB
- Power draw
- 18.1 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-4B
- Quantization
- FP16
- Framework
- vLLM
- Version
- v0.23
- 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 M1 Max
- Measured VRAM (avg)
- 17.64 GB
- OS
- macOS 15.0
- Driver
- Apple M1 Max
- CUDA
- N/A
- Power draw
- 18.1 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.23 · macOS 15.0 · CUDA N/A · 16384 ctx
57.12 tok/s
Decode
260.15 tok/s
Prefill
1.28 s
TTFT
17.64 GB
VRAM
—
MTP
18.1 W
Power draw
L2 batch seed 2026-10-16: vLLM on hw_id=101 running model_id=48 at FP16