Arao AI
Back to search results

Qwen3.5-27B · IQ2_M

Start a discussion
NVIDIA RTX 509032GBOllama v0.37L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-27B (IQ2_M) on NVIDIA RTX 5090 with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.

235.26 tok/s

Decode

Decode speed

1,229.12 tok/s

Prefill

Prefill speed

2.25 s

TTFT

Time to first token

30.4 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. Ollama · IQ1_S Decode 260.56 · Prefill 868.47 ·
  2. Ollama · IQ2_M (current)Decode 235.26 · Prefill 1229.12 ·
  3. llama.cpp · IQ4_XS Decode 134 · Prefill 1478 ·
  4. llama.cpp · Q5_K_M Decode 114 · Prefill 1331 ·
  5. llama.cpp · Q8_0 Decode 92 · Prefill 792 ·
  6. vLLM · IQ4_XS Decode 108 · Prefill 1288 ·
  7. vLLM · Q4_K_M Decode 100 · Prefill 1182 ·
  8. vLLM · Q5_K_M Decode 106 · Prefill 956 ·
  9. vLLM · Q8_0 Decode 74 · Prefill 795 ·
  10. Ollama · IQ4_XS Decode 136 · Prefill 1332 ·
  11. Ollama · Q4_K_M Decode 125 · Prefill 1177 ·
  12. Ollama · Q5_K_M Decode 113 · Prefill 1178 ·
  13. Ollama · Q8_0 Decode 75 · Prefill 899 ·
  14. Strata · IQ4_XS Decode 198 · Prefill 1923 ·
  15. Strata · Q4_K_M Decode 178 · Prefill 1794 ·
  16. Strata · Q5_K_M Decode 147 · Prefill 1764 ·
  17. Strata · Q8_0 Decode 127 · Prefill 1290 ·
  18. llama.cpp · Q4_K_M Decode 30.09 · Prefill 1515.83 ·

Core figures

Decode (avg)
235.26 tok/s
Prefill (avg)
1,229.12 tok/s
TTFT (avg)
2.25 s
VRAM (avg)
30.4 GB
MTP acceptance rate
—
TTFB
1.88 GB
Power draw
497.4 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-27B
Quantization
IQ2_M
Framework
Ollama
Version
v0.37
Context length
32768 tokens
Batch size
256
GPU layers
99
Flash Attention
On

Hardware

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

GPU
NVIDIA RTX 5090
Nominal VRAM
32 GB
Measured VRAM (avg)
30.4 GB
OS
Windows 11 23H2
Driver
555.85
CUDA
12.6
Power draw
497.4 W

Sources & evidence

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

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

    v0.37 · Windows 11 23H2 · CUDA 12.6 · 32768 ctx

    235.26 tok/s

    Decode

    1,229.12 tok/s

    Prefill

    2.25 s

    TTFT

    30.4 GB

    VRAM

    —

    MTP

    497.4 W

    Power draw

    L2 batch seed 2026-10-10: Ollama on hw_id=13 running model_id=50 at IQ2_M