Arao AI
Back to search results

Flint-40B-A2.5B · IQ2_M

Start a discussion
NVIDIA H100 80GB HBM380GBvLLM v0.21L2 Cross-framework verified

This page aggregates 1 real-world runs of Flint-40B-A2.5B (IQ2_M) on NVIDIA H100 80GB HBM3 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

325.98 tok/s

Decode

Decode speed

1,095.54 tok/s

Prefill

Prefill speed

2.24 s

TTFT

Time to first token

45.56 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

14 days ago

Last verified

Performance

  1. vLLM · IQ2_M (current)Decode 325.98 · Prefill 1095.54 ·

Core figures

Decode (avg)
325.98 tok/s
Prefill (avg)
1,095.54 tok/s
TTFT (avg)
2.24 s
VRAM (avg)
45.56 GB
MTP acceptance rate
—
TTFB
2.61 GB
Power draw
690.6 W

Configuration

Member-level fields are taken from the most recent run

Model
Flint-40B-A2.5B
Quantization
IQ2_M
Framework
vLLM
Version
v0.21
Context length
32768 tokens
Batch size
32
GPU layers
999
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)
45.56 GB
OS
Ubuntu 22.04
Driver
560.94
CUDA
13.0
Power draw
690.6 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

    v0.21 · Ubuntu 22.04 · CUDA 13.0 · 32768 ctx

    325.98 tok/s

    Decode

    1,095.54 tok/s

    Prefill

    2.24 s

    TTFT

    45.56 GB

    VRAM

    —

    MTP

    690.6 W

    Power draw

    L2 batch seed 2026-10-22: vLLM on hw_id=55 running model_id=92 at IQ2_M