gemma-4-31B-it · FP16
This page aggregates 1 real-world runs of gemma-4-31B-it (FP16) on NVIDIA RTX Spark with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
4.66 tok/s
Decode
Decode speed
23.51 tok/s
Prefill
Prefill speed
3.09 s
TTFT
Time to first token
110.2 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
16 days ago
Last verified
Performance
- llama.cpp · FP16 (current)Decode 4.66 · Prefill 23.51 ·
- llama.cpp · Q6_K Decode 6.37 · Prefill 31.53 ·
- llama.cpp · Q8_0 Decode 5.75 · Prefill 29.94 ·
- Ollama · FP16 Decode 3.97 · Prefill 19.3 ·
- Ollama · Q8_0 Decode 5.28 · Prefill 19.56 ·
- llama.cpp · Q4_K_M Decode 8.84 · Prefill 36.97 ·
- Ollama · Q3_K_S Decode 10.24 · Prefill 41.22 ·
- Ollama · Q4_K_M Decode 8 · Prefill 37.56 ·
Core figures
- Decode (avg)
- 4.66 tok/s
- Prefill (avg)
- 23.51 tok/s
- TTFT (avg)
- 3.09 s
- VRAM (avg)
- 110.2 GB
- MTP acceptance rate
- —
- TTFB
- 0.04 GB
- Power draw
- 77.5 W
Configuration
Member-level fields are taken from the most recent run
- Model
- gemma-4-31B-it
- Quantization
- FP16
- Framework
- llama.cpp
- Version
- v0.27
- Context length
- 32768 tokens
- Batch size
- 256
- GPU layers
- 33
- Flash Attention
- Off
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX Spark
- Nominal VRAM
- 128 GB
- Measured VRAM (avg)
- 110.2 GB
- OS
- Ubuntu 24.04
- Driver
- 615.78
- CUDA
- 13.0
- Power draw
- 77.5 W
- Reference price
- $2,399 · updated 2026-10-19
Prices are not live; check the store page
Sources & evidence
1 measured records in total, each traceable to its original source
v0.27 · Ubuntu 24.04 · CUDA 13.0 · 32768 ctx
4.66 tok/s
Decode
23.51 tok/s
Prefill
3.09 s
TTFT
110.2 GB
VRAM
—
MTP
77.5 W
Power draw
L2 batch seed 2026-10-20 (Linux 116GiB unlock): llama.cpp on Spark running model_id=63 at FP16