Arao AI
Back to search results

Qwen3.5-4B · IQ4_XS

Start a discussion
Intel Arc B58012GBllama.cpp v0.53L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-4B (IQ4_XS) on Intel Arc B580 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

64.83 tok/s

Decode

Decode speed

327.91 tok/s

Prefill

Prefill speed

2.27 s

TTFT

Time to first token

8.32 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

16 days ago

Last verified

Performance

  1. llama.cpp · IQ2_XS Decode 84.02 · Prefill 284.07 ·
  2. llama.cpp · IQ4_XS (current)Decode 64.83 · Prefill 327.91 ·
  3. Ollama · IQ2_M Decode 78.85 · Prefill 313.28 ·
  4. Ollama · Q8_0 Decode 37.78 · Prefill 141.17 ·
  5. llama.cpp · Q3_K_S Decode 74.37 · Prefill 373.85 ·
  6. Strata · Q6_K Decode 60.78 · Prefill 199.46 ·
  7. llama.cpp · IQ1_S Decode 98.24 · Prefill 506.66 ·
  8. llama.cpp · Q3_K_L Decode 70.52 · Prefill 317.93 ·
  9. vLLM · Q2_K Decode 89.29 · Prefill 348.31 ·
  10. Strata · IQ3_S Decode 101.32 · Prefill 490.59 ·

Core figures

Decode (avg)
64.83 tok/s
Prefill (avg)
327.91 tok/s
TTFT (avg)
2.27 s
VRAM (avg)
8.32 GB
MTP acceptance rate
—
TTFB
0.52 GB
Power draw
160.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-4B
Quantization
IQ4_XS
Framework
llama.cpp
Version
v0.53
Context length
2048 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
Intel Arc B580
Nominal VRAM
12 GB
Measured VRAM (avg)
8.32 GB
OS
Windows 11 24H2
Driver
Intel Arc 32.0.101.6552
CUDA
12.4
Power draw
160.8 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-19

    v0.53 · Windows 11 24H2 · CUDA 12.4 · 2048 ctx

    64.83 tok/s

    Decode

    327.91 tok/s

    Prefill

    2.27 s

    TTFT

    8.32 GB

    VRAM

    —

    MTP

    160.8 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=94 running model_id=48 at IQ4_XS