Arao AI
Back to search results

Qwen3.5-0.8B · Q4_K_M

Start a discussion
NVIDIA RTX 4080 Super16GBllama.cpp v0.36L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-0.8B (Q4_K_M) on NVIDIA RTX 4080 Super with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

344.73 tok/s

Decode

Decode speed

1,737.55 tok/s

Prefill

Prefill speed

0.87 s

TTFT

Time to first token

3.48 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. llama.cpp · Q4_K_M (current)Decode 344.73 · Prefill 1737.55 ·
  2. vLLM · IQ1_S Decode 314.79 · Prefill 1409.84 ·

Core figures

Decode (avg)
344.73 tok/s
Prefill (avg)
1,737.55 tok/s
TTFT (avg)
0.87 s
VRAM (avg)
3.48 GB
MTP acceptance rate
—
TTFB
2.76 GB
Power draw
255.1 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-0.8B
Quantization
Q4_K_M
Framework
llama.cpp
Version
v0.36
Context length
2048 tokens
Batch size
256
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 Super
Nominal VRAM
16 GB
Measured VRAM (avg)
3.48 GB
OS
Ubuntu 22.04
Driver
560.94
CUDA
12.8
Power draw
255.1 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

    v0.36 · Ubuntu 22.04 · CUDA 12.8 · 2048 ctx

    344.73 tok/s

    Decode

    1,737.55 tok/s

    Prefill

    0.87 s

    TTFT

    3.48 GB

    VRAM

    —

    MTP

    255.1 W

    Power draw

    L2 batch seed 2026-10-18: llama.cpp on hw_id=20 running model_id=51 at Q4_K_M