Arao AI
Back to search results

gemma-4-12B-it · IQ3_XS

Start a discussion
NVIDIA RTX 5070 Ti16GBOllama v0.21L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-12B-it (IQ3_XS) on NVIDIA RTX 5070 Ti with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.

80.61 tok/s

Decode

Decode speed

400.97 tok/s

Prefill

Prefill speed

0.75 s

TTFT

Time to first token

15.2 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. Ollama · IQ3_XS (current)Decode 80.61 · Prefill 400.97 ·
  2. vLLM · Q8_0 Decode 58.6 · Prefill 439.5 ·
  3. vLLM · Q5_K_M Decode 85.6 · Prefill 642 ·
  4. vLLM · Q4_K_M Decode 104.3 · Prefill 782.3 ·
  5. llama.cpp · Q8_0 Decode 53.8 · Prefill 403.5 ·
  6. llama.cpp · Q5_K_M Decode 78.5 · Prefill 588.8 ·
  7. llama.cpp · Q4_K_M Decode 95.7 · Prefill 717.8 ·

Core figures

Decode (avg)
80.61 tok/s
Prefill (avg)
400.97 tok/s
TTFT (avg)
0.75 s
VRAM (avg)
15.2 GB
MTP acceptance rate
—
TTFB
0.64 GB
Power draw
266.4 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-12B-it
Quantization
IQ3_XS
Framework
Ollama
Version
v0.21
Context length
32768 tokens
Batch size
512
GPU layers
28
Flash Attention
On

Hardware

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

GPU
NVIDIA RTX 5070 Ti
Nominal VRAM
16 GB
Measured VRAM (avg)
15.2 GB
OS
Windows 11 23H2
Driver
560.94
CUDA
12.6
Power draw
266.4 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-18

    v0.21 · Windows 11 23H2 · CUDA 12.6 · 32768 ctx

    80.61 tok/s

    Decode

    400.97 tok/s

    Prefill

    0.75 s

    TTFT

    15.2 GB

    VRAM

    —

    MTP

    266.4 W

    Power draw

    L2 batch seed 2026-10-18: Ollama on hw_id=17 running model_id=65 at IQ3_XS