Back to search results
Qwen3-4B-Thinking-2507 · IQ4_XS
Apple M4 Pro64GBllama.cpp v0.45L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3-4B-Thinking-2507 (IQ4_XS) on Apple M4 Pro with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
138.37 tok/s
Decode
Decode speed
661.86 tok/s
Prefill
Prefill speed
0.46 s
TTFT
Time to first token
8.32 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
5 days ago
Last verified
Performance
- llama.cpp · IQ4_XS (current)Decode 138.37 · Prefill 661.86 ·
- llama.cpp · Q4_K_S Decode 123.54 · Prefill 441.3 ·
- Strata · IQ3_XS Decode 218.39 · Prefill 1145.06 ·
- Strata · Q3_K_L Decode 206.28 · Prefill 783.16 ·
Core figures
- Decode (avg)
- 138.37 tok/s
- Prefill (avg)
- 661.86 tok/s
- TTFT (avg)
- 0.46 s
- VRAM (avg)
- 8.32 GB
- MTP acceptance rate
- —
- TTFB
- 1.11 GB
- Power draw
- 40.3 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3-4B-Thinking-2507
- Quantization
- IQ4_XS
- Framework
- llama.cpp
- Version
- v0.45
- Context length
- 8192 tokens
- Batch size
- 512
- GPU layers
- 28
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- Apple M4 Pro
- Nominal VRAM
- 64 GB
- Measured VRAM (avg)
- 8.32 GB
- OS
- macOS 15.0
- Driver
- Apple M4 Pro
- CUDA
- N/A
- Power draw
- 40.3 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.45 · macOS 15.0 · CUDA N/A · 8192 ctx
138.37 tok/s
Decode
661.86 tok/s
Prefill
0.46 s
TTFT
8.32 GB
VRAM
—
MTP
40.3 W
Power draw
L2 batch seed 2026-10-13: llama.cpp on hw_id=25 running model_id=57 at IQ4_XS