Back to search results
gemma-4-12B-it · IQ3_XS
Apple M4 Max128GBvLLM v0.23L2 Cross-framework verified
This page aggregates 1 real-world runs of gemma-4-12B-it (IQ3_XS) on Apple M4 Max with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
181.71 tok/s
Decode
Decode speed
766.49 tok/s
Prefill
Prefill speed
0.36 s
TTFT
Time to first token
17.72 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · IQ2_M Decode 187.13 · Prefill 653.55 ·
- llama.cpp · Q4_K_M Decode 140.93 · Prefill 615.86 ·
- vLLM · IQ3_XS (current)Decode 181.71 · Prefill 766.49 ·
- Strata · Q2_K Decode 270.81 · Prefill 1355.35 ·
Core figures
- Decode (avg)
- 181.71 tok/s
- Prefill (avg)
- 766.49 tok/s
- TTFT (avg)
- 0.36 s
- VRAM (avg)
- 17.72 GB
- MTP acceptance rate
- —
- TTFB
- 1.45 GB
- Power draw
- 70.3 W
Configuration
Member-level fields are taken from the most recent run
- Model
- gemma-4-12B-it
- Quantization
- IQ3_XS
- Framework
- vLLM
- Version
- v0.23
- Context length
- 8192 tokens
- Batch size
- 16
- 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)
- 17.72 GB
- OS
- macOS 15.0
- Driver
- Apple M4 Max
- CUDA
- N/A
- Power draw
- 70.3 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.23 · macOS 15.0 · CUDA N/A · 8192 ctx
181.71 tok/s
Decode
766.49 tok/s
Prefill
0.36 s
TTFT
17.72 GB
VRAM
—
MTP
70.3 W
Power draw
L2 batch seed 2026-10-10: vLLM on hw_id=26 running model_id=65 at IQ3_XS