Qwen3.8-27B · IQ3_XS
This page aggregates 1 real-world runs of Qwen3.8-27B (IQ3_XS) on Apple M3 Max with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
96.32 tok/s
Decode
Decode speed
353.29 tok/s
Prefill
Prefill speed
0.51 s
TTFT
Time to first token
34.2 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · IQ3_XS (current)Decode 96.32 · Prefill 353.29 ·
- llama.cpp · IQ2_XS Decode 25.7 · Prefill 192.8 ·
- llama.cpp · Q5_K_M Decode 11.7 · Prefill 87.8 ·
- llama.cpp · Q4_K_M Decode 14.3 · Prefill 107.3 ·
Core figures
- Decode (avg)
- 96.32 tok/s
- Prefill (avg)
- 353.29 tok/s
- TTFT (avg)
- 0.51 s
- VRAM (avg)
- 34.2 GB
- MTP acceptance rate
- —
- TTFB
- 0.77 GB
- Power draw
- 88.7 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.8-27B
- Quantization
- IQ3_XS
- Framework
- llama.cpp
- Version
- v0.26
- Context length
- 16384 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 M3 Max
- Nominal VRAM
- 36 GB
- Measured VRAM (avg)
- 34.2 GB
- OS
- macOS 15.0
- Driver
- Apple M3 Max
- CUDA
- N/A
- Power draw
- 88.7 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
96.32 tok/s
Decode
353.29 tok/s
Prefill
0.51 s
TTFT
34.2 GB
VRAM
—
MTP
88.7 W
Power draw
L2 batch seed 2026-10-10: llama.cpp on hw_id=4 running model_id=5 at IQ3_XS