Back to search results
Qwen3-1.7B · Q4_K_M
Apple M2 Ultrallama.cpp v0.39L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3-1.7B (Q4_K_M) on Apple M2 Ultra with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
200.31 tok/s
Decode
Decode speed
817.12 tok/s
Prefill
Prefill speed
2.06 s
TTFT
Time to first token
5.03 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
2 days ago
Last verified
Performance
- llama.cpp · Q2_K Decode 247.88 · Prefill 1258.27 ·
- llama.cpp · Q4_K_M (current)Decode 200.31 · Prefill 817.12 ·
- vLLM · IQ4_XS Decode 221.65 · Prefill 888.85 ·
- vLLM · Q6_K Decode 162.21 · Prefill 522.8 ·
Core figures
- Decode (avg)
- 200.31 tok/s
- Prefill (avg)
- 817.12 tok/s
- TTFT (avg)
- 2.06 s
- VRAM (avg)
- 5.03 GB
- MTP acceptance rate
- —
- TTFB
- 1.6 GB
- Power draw
- 62.1 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3-1.7B
- Quantization
- Q4_K_M
- Framework
- llama.cpp
- Version
- v0.39
- Context length
- 2048 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)
- 5.03 GB
- OS
- macOS 15.0
- Driver
- Apple M2 Ultra
- CUDA
- N/A
- Power draw
- 62.1 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.39 · macOS 15.0 · CUDA N/A · 2048 ctx
200.31 tok/s
Decode
817.12 tok/s
Prefill
2.06 s
TTFT
5.03 GB
VRAM
—
MTP
62.1 W
Power draw
L2 batch seed 2026-10-16: llama.cpp on hw_id=103 running model_id=55 at Q4_K_M