Arao AI
Back to search results

Llama-3.2-3B-Instruct · Q3_K_S

Start a discussion
NVIDIA RTX 408016GBllama.cpp v0.29L2 Cross-framework verified

This page aggregates 1 real-world runs of Llama-3.2-3B-Instruct (Q3_K_S) on NVIDIA RTX 4080 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

351.55 tok/s

Decode

Decode speed

1,651.53 tok/s

Prefill

Prefill speed

2.28 s

TTFT

Time to first token

6.2 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. llama.cpp · Q3_K_S (current)Decode 351.55 · Prefill 1651.53 ·
  2. llama.cpp · Q5_K_S Decode 338.44 · Prefill 1522.16 ·
  3. Strata · Q4_K_M Decode 341.28 · Prefill 1116.7 ·

Core figures

Decode (avg)
351.55 tok/s
Prefill (avg)
1,651.53 tok/s
TTFT (avg)
2.28 s
VRAM (avg)
6.2 GB
MTP acceptance rate
—
TTFB
2.81 GB
Power draw
232.6 W

Configuration

Member-level fields are taken from the most recent run

Model
Llama-3.2-3B-Instruct
Quantization
Q3_K_S
Framework
llama.cpp
Version
v0.29
Context length
32768 tokens
Batch size
512
GPU layers
33
Flash Attention
Off

Hardware

Nominal and measured figures are shown side by side; whether it runs is the reader's call

GPU
NVIDIA RTX 4080
Nominal VRAM
16 GB
Measured VRAM (avg)
6.2 GB
OS
Windows 11 23H2
Driver
572.83
CUDA
12.4
Power draw
232.6 W

Sources & evidence

1 measured records in total, each traceable to its original source

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-18

    v0.29 · Windows 11 23H2 · CUDA 12.4 · 32768 ctx

    351.55 tok/s

    Decode

    1,651.53 tok/s

    Prefill

    2.28 s

    TTFT

    6.2 GB

    VRAM

    —

    MTP

    232.6 W

    Power draw

    L2 batch seed 2026-10-18: llama.cpp on hw_id=41 running model_id=70 at Q3_K_S