Qwen3.8-27B · IQ2_XS
This page aggregates 1 real-world runs of Qwen3.8-27B (IQ2_XS) on Apple M2 Ultra with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
77.57 tok/s
Decode
Decode speed
340.43 tok/s
Prefill
Prefill speed
0.89 s
TTFT
Time to first token
29.32 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
17 days ago
Last verified
Performance
- llama.cpp · IQ2_XS (current)Decode 77.57 · Prefill 340.43 ·
- Ollama · IQ2_XS Decode 73.6 · Prefill 261.9 ·
- Ollama · IQ3_XS Decode 64.55 · Prefill 256.75 ·
Core figures
- Decode (avg)
- 77.57 tok/s
- Prefill (avg)
- 340.43 tok/s
- TTFT (avg)
- 0.89 s
- VRAM (avg)
- 29.32 GB
- MTP acceptance rate
- —
- TTFB
- 0.62 GB
- Power draw
- 129.4 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.8-27B
- Quantization
- IQ2_XS
- Framework
- llama.cpp
- Version
- v0.22
- Context length
- 4096 tokens
- Batch size
- 256
- GPU layers
- 99
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- Apple M2 Ultra
- Measured VRAM (avg)
- 29.32 GB
- OS
- macOS 26.1
- Power draw
- 129.4 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
v0.22 · macOS 26.1 · 4096 ctx
77.57 tok/s
Decode
340.43 tok/s
Prefill
0.89 s
TTFT
29.32 GB
VRAM
—
MTP
129.4 W
Power draw
L2 batch seed 2026-10-19: llama.cpp on hw_id=103 running model_id=5 at IQ2_XS