Arao AI
Back to search results

Qwen3-1.7B · Q8_0

Start a discussion
Apple M5 Max128GBStrata v0.25L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-1.7B (Q8_0) on Apple M5 Max with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

319.85 tok/s

Decode

Decode speed

1,662.65 tok/s

Prefill

Prefill speed

0.81 s

TTFT

Time to first token

6.21 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. vLLM · Q2_K Decode 302.27 · Prefill 1354.5 ·
  2. vLLM · Q8_0 Decode 255.36 · Prefill 1217.53 ·
  3. Strata · Q8_0 (current)Decode 319.85 · Prefill 1662.65 ·
  4. Ollama · Q3_K_L Decode 353.08 · Prefill 1662.13 ·

Core figures

Decode (avg)
319.85 tok/s
Prefill (avg)
1,662.65 tok/s
TTFT (avg)
0.81 s
VRAM (avg)
6.21 GB
MTP acceptance rate
0.62
TTFB
2.56 GB
Power draw
278.7 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-1.7B
Quantization
Q8_0
Framework
Strata
Version
v0.25
Context length
4096 tokens
Batch size
512
GPU layers
5
Flash Attention
Off

Hardware

Nominal and measured figures are shown side by side; whether it runs is the reader's call

GPU
Apple M5 Max
Nominal VRAM
128 GB
Measured VRAM (avg)
6.21 GB
OS
Ubuntu 24.04
Driver
565.57
CUDA
12.8
Power draw
278.7 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-16

    v0.25 · Ubuntu 24.04 · CUDA 12.8 · 4096 ctx

    319.85 tok/s

    Decode

    1,662.65 tok/s

    Prefill

    0.81 s

    TTFT

    6.21 GB

    VRAM

    0.62

    MTP

    278.7 W

    Power draw

    L2 batch seed 2026-10-16: Strata on hw_id=28 running model_id=55 at Q8_0