Arao AI
Back to search results

Ministral-3-8B-Instruct-2512 · Q8_0

Start a discussion
NVIDIA H100 80GB HBM380GBllama.cpp v0.37L2 Cross-framework verified

This page aggregates 1 real-world runs of Ministral-3-8B-Instruct-2512 (Q8_0) on NVIDIA H100 80GB HBM3 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

271.67 tok/s

Decode

Decode speed

1,471.82 tok/s

Prefill

Prefill speed

0.64 s

TTFT

Time to first token

21.42 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

12 days ago

Last verified

Performance

  1. llama.cpp · Q8_0 (current)Decode 271.67 · Prefill 1471.82 ·
  2. vLLM · Q3_K_M Decode 321.73 · Prefill 1464.84 ·

Core figures

Decode (avg)
271.67 tok/s
Prefill (avg)
1,471.82 tok/s
TTFT (avg)
0.64 s
VRAM (avg)
21.42 GB
MTP acceptance rate
—
TTFB
2.17 GB
Power draw
621.6 W

Configuration

Member-level fields are taken from the most recent run

Model
Ministral-3-8B-Instruct-2512
Quantization
Q8_0
Framework
llama.cpp
Version
v0.37
Context length
65536 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 H100 80GB HBM3
Nominal VRAM
80 GB
Measured VRAM (avg)
21.42 GB
OS
Ubuntu 24.04
Driver
565.57
CUDA
13.0
Power draw
621.6 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-24

    v0.37 · Ubuntu 24.04 · CUDA 13.0 · 65536 ctx

    271.67 tok/s

    Decode

    1,471.82 tok/s

    Prefill

    0.64 s

    TTFT

    21.42 GB

    VRAM

    —

    MTP

    621.6 W

    Power draw

    L2 batch seed 2026-10-24: llama.cpp on hw_id=55 running model_id=74 at Q8_0