Back to search results
Qwen3-4B-Thinking-2507 · IQ2_M
Apple M2 UltravLLM v0.22L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3-4B-Thinking-2507 (IQ2_M) on Apple M2 Ultra with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
191.2 tok/s
Decode
Decode speed
838.86 tok/s
Prefill
Prefill speed
2.06 s
TTFT
Time to first token
6.47 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
2 days ago
Last verified
Performance
- vLLM · IQ2_M (current)Decode 191.2 · Prefill 838.86 ·
- Ollama · Q3_K_M Decode 131.4 · Prefill 507.32 ·
Core figures
- Decode (avg)
- 191.2 tok/s
- Prefill (avg)
- 838.86 tok/s
- TTFT (avg)
- 2.06 s
- VRAM (avg)
- 6.47 GB
- MTP acceptance rate
- —
- TTFB
- 1.53 GB
- Power draw
- 43.3 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3-4B-Thinking-2507
- Quantization
- IQ2_M
- Framework
- vLLM
- Version
- v0.22
- Context length
- 4096 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 M2 Ultra
- Measured VRAM (avg)
- 6.47 GB
- OS
- macOS 15.0
- Driver
- Apple M2 Ultra
- CUDA
- N/A
- Power draw
- 43.3 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.22 · macOS 15.0 · CUDA N/A · 4096 ctx
191.2 tok/s
Decode
838.86 tok/s
Prefill
2.06 s
TTFT
6.47 GB
VRAM
—
MTP
43.3 W
Power draw
L2 batch seed 2026-10-16: vLLM on hw_id=103 running model_id=57 at IQ2_M