Back to search results
Kimi-K3 · Q4_K_M
NVIDIA RTX 309024GBllama.cpp v0.28L2 Cross-framework verified
This page aggregates 1 real-world runs of Kimi-K3 (Q4_K_M) on NVIDIA RTX 3090 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
61.22 tok/s
Decode
Decode speed
289.69 tok/s
Prefill
Prefill speed
1.04 s
TTFT
Time to first token
1.05 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · Q4_K_M (current)Decode 61.22 · Prefill 289.69 ·
- Ollama · Q5_K_M Decode 54.04 · Prefill 184.49 ·
Core figures
- Decode (avg)
- 61.22 tok/s
- Prefill (avg)
- 289.69 tok/s
- TTFT (avg)
- 1.04 s
- VRAM (avg)
- 1.05 GB
- MTP acceptance rate
- —
- TTFB
- 0.49 GB
- Power draw
- 300.5 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Kimi-K3
- Quantization
- Q4_K_M
- Framework
- llama.cpp
- Version
- v0.28
- Context length
- 32768 tokens
- Batch size
- 256
- 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
- NVIDIA RTX 3090
- Nominal VRAM
- 24 GB
- Measured VRAM (avg)
- 1.05 GB
- OS
- Windows 11 23H2
- Driver
- 572.83
- CUDA
- 12.4
- Power draw
- 300.5 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.28 · Windows 11 23H2 · CUDA 12.4 · 32768 ctx
61.22 tok/s
Decode
289.69 tok/s
Prefill
1.04 s
TTFT
1.05 GB
VRAM
—
MTP
300.5 W
Power draw
L2 batch seed 2026-10-18: llama.cpp on hw_id=2 running model_id=73 at Q4_K_M