Arao AI
Back to search results

mammoth-1 · Q3_K_S

Start a discussion
NVIDIA H100 80GB HBM380GBStrata v0.28L2 Cross-framework verified

This page aggregates 1 real-world runs of mammoth-1 (Q3_K_S) on NVIDIA H100 80GB HBM3 with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

345.96 tok/s

Decode

Decode speed

1,251.15 tok/s

Prefill

Prefill speed

1.63 s

TTFT

Time to first token

38.96 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

13 days ago

Last verified

Performance

  1. llama.cpp · Q4_K_S Decode 342.54 · Prefill 1113.53 ·
  2. Strata · Q3_K_S (current)Decode 345.96 · Prefill 1251.15 ·

Core figures

Decode (avg)
345.96 tok/s
Prefill (avg)
1,251.15 tok/s
TTFT (avg)
1.63 s
VRAM (avg)
38.96 GB
MTP acceptance rate
0.47
TTFB
2.77 GB
Power draw
694.6 W

Configuration

Member-level fields are taken from the most recent run

Model
mammoth-1
Quantization
Q3_K_S
Framework
Strata
Version
v0.28
Context length
4096 tokens
Batch size
1024
GPU layers
8
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)
38.96 GB
OS
Ubuntu 22.04
Driver
572.83
CUDA
13.0
Power draw
694.6 W

Get started

Original model resources

Base model resources may not match this quantization. Use the verified deployment weights above when available.

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-22

    v0.28 · Ubuntu 22.04 · CUDA 13.0 · 4096 ctx

    345.96 tok/s

    Decode

    1,251.15 tok/s

    Prefill

    1.63 s

    TTFT

    38.96 GB

    VRAM

    0.47

    MTP

    694.6 W

    Power draw

    L2 batch seed 2026-10-22: Strata on hw_id=55 running model_id=94 at Q3_K_S