Arao AI
Back to search results

Llama-3.1-8B-Instruct · Q3_K_L

Start a discussion
NVIDIA RTX 408016GBllama.cpp v0.33L2 Cross-framework verified

This page aggregates 1 real-world runs of Llama-3.1-8B-Instruct (Q3_K_L) on NVIDIA RTX 4080 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

160.83 tok/s

Decode

Decode speed

637.28 tok/s

Prefill

Prefill speed

0.78 s

TTFT

Time to first token

14.2 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_L (current)Decode 160.83 · Prefill 637.28 ·
  2. Strata · IQ4_XS Decode 248.44 · Prefill 973.3 ·
  3. Strata · Q3_K_M Decode 238.79 · Prefill 1287.02 ·

Core figures

Decode (avg)
160.83 tok/s
Prefill (avg)
637.28 tok/s
TTFT (avg)
0.78 s
VRAM (avg)
14.2 GB
MTP acceptance rate
—
TTFB
1.29 GB
Power draw
298.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Llama-3.1-8B-Instruct
Quantization
Q3_K_L
Framework
llama.cpp
Version
v0.33
Context length
2048 tokens
Batch size
256
GPU layers
20
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)
14.2 GB
OS
Ubuntu 22.04
Driver
572.83
CUDA
12.4
Power draw
298.8 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

    v0.33 · Ubuntu 22.04 · CUDA 12.4 · 2048 ctx

    160.83 tok/s

    Decode

    637.28 tok/s

    Prefill

    0.78 s

    TTFT

    14.2 GB

    VRAM

    —

    MTP

    298.8 W

    Power draw

    L2 batch seed 2026-10-18: llama.cpp on hw_id=41 running model_id=68 at Q3_K_L