elephant-1 · Q4_K_M
This page aggregates 1 real-world runs of elephant-1 (Q4_K_M) on Apple M2 Ultra with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
92.52 tok/s
Decode
Decode speed
377.54 tok/s
Prefill
Prefill speed
2.38 s
TTFT
Time to first token
17.6 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
13 days ago
Last verified
Performance
- llama.cpp · Q4_K_M (current)Decode 92.52 · Prefill 377.54 ·
- llama.cpp · Q4_K_S Decode 88.6 · Prefill 484.39 ·
- Ollama · Q6_K Decode 67.84 · Prefill 354.32 ·
Core figures
- Decode (avg)
- 92.52 tok/s
- Prefill (avg)
- 377.54 tok/s
- TTFT (avg)
- 2.38 s
- VRAM (avg)
- 17.6 GB
- MTP acceptance rate
- —
- TTFB
- 0.74 GB
- Power draw
- 110.5 W
Configuration
Member-level fields are taken from the most recent run
- Model
- elephant-1
- Quantization
- Q4_K_M
- Framework
- llama.cpp
- Version
- v0.38
- Context length
- 16384 tokens
- Batch size
- 256
- GPU layers
- 99
- Flash Attention
- Off
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)
- 17.6 GB
- OS
- macOS 26.1
- Power draw
- 110.5 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
v0.38 · macOS 26.1 · 16384 ctx
92.52 tok/s
Decode
377.54 tok/s
Prefill
2.38 s
TTFT
17.6 GB
VRAM
—
MTP
110.5 W
Power draw
L2 batch seed 2026-10-22: llama.cpp on hw_id=103 running model_id=95 at Q4_K_M