Back to search results
Llama-3.2-3B-Instruct · Q4_K_S
NVIDIA RTX 409024GBOllama v0.45L2 Cross-framework verified
This page aggregates 1 real-world runs of Llama-3.2-3B-Instruct (Q4_K_S) on NVIDIA RTX 4090 with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.
336.75 tok/s
Decode
Decode speed
1,261.4 tok/s
Prefill
Prefill speed
1.01 s
TTFT
Time to first token
7.01 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
16 days ago
Last verified
Performance
- vLLM · IQ3_S Decode 327.57 · Prefill 1515.42 ·
- Ollama · Q4_K_S (current)Decode 336.75 · Prefill 1261.4 ·
- Ollama · Q5_K_M Decode 263.12 · Prefill 1197.54 ·
- Strata · IQ1_S Decode 284.1 · Prefill 1178.93 ·
- Strata · IQ2_XS Decode 281.21 · Prefill 1389.98 ·
- llama.cpp · Q8_0 Decode 245.7 · Prefill 1842.8 ·
- llama.cpp · Q5_K_M Decode 357.4 · Prefill 2680.5 ·
- llama.cpp · Q4_K_M Decode 436.8 · Prefill 3276 ·
Core figures
- Decode (avg)
- 336.75 tok/s
- Prefill (avg)
- 1,261.4 tok/s
- TTFT (avg)
- 1.01 s
- VRAM (avg)
- 7.01 GB
- MTP acceptance rate
- —
- TTFB
- 2.69 GB
- Power draw
- 361.5 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Llama-3.2-3B-Instruct
- Quantization
- Q4_K_S
- Framework
- Ollama
- Version
- v0.45
- Context length
- 4096 tokens
- Batch size
- 256
- GPU layers
- 33
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 4090
- Nominal VRAM
- 24 GB
- Measured VRAM (avg)
- 7.01 GB
- OS
- Ubuntu 24.04
- Driver
- 572.83
- CUDA
- 12.6
- Power draw
- 361.5 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.45 · Ubuntu 24.04 · CUDA 12.6 · 4096 ctx
336.75 tok/s
Decode
1,261.4 tok/s
Prefill
1.01 s
TTFT
7.01 GB
VRAM
—
MTP
361.5 W
Power draw
L2 batch seed 2026-10-19: Ollama on hw_id=1 running model_id=70 at Q4_K_S