Back to search results
Qwen3.8-9B · IQ3_S
NVIDIA RTX 508016GBllama.cpp v0.25L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.8-9B (IQ3_S) on NVIDIA RTX 5080 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
200.28 tok/s
Decode
Decode speed
716.67 tok/s
Prefill
Prefill speed
1.97 s
TTFT
Time to first token
14.57 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
17 days ago
Last verified
Performance
- llama.cpp · IQ3_S (current)Decode 200.28 · Prefill 716.67 ·
- vLLM · Q2_K Decode 238.4 · Prefill 951.3 ·
- vLLM · Q4_K_M Decode 152.58 · Prefill 685.84 ·
- Strata · IQ2_M Decode 294.71 · Prefill 1507.15 ·
- Strata · IQ3_XS Decode 265.22 · Prefill 1010.61 ·
- llama.cpp · IQ4_XS Decode 97.14 · Prefill 496.38 ·
- llama.cpp · Q3_K_M Decode 108.92 · Prefill 455.22 ·
- vLLM · IQ2_M Decode 131.31 · Prefill 559.09 ·
Core figures
- Decode (avg)
- 200.28 tok/s
- Prefill (avg)
- 716.67 tok/s
- TTFT (avg)
- 1.97 s
- VRAM (avg)
- 14.57 GB
- MTP acceptance rate
- —
- TTFB
- 1.6 GB
- Power draw
- 282.3 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.8-9B
- Quantization
- IQ3_S
- Framework
- llama.cpp
- Version
- v0.25
- Context length
- 8192 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 5080
- Nominal VRAM
- 16 GB
- Measured VRAM (avg)
- 14.57 GB
- OS
- Windows 11 23H2
- Driver
- 565.57
- CUDA
- 12.6
- Power draw
- 282.3 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.25 · Windows 11 23H2 · CUDA 12.6 · 8192 ctx
200.28 tok/s
Decode
716.67 tok/s
Prefill
1.97 s
TTFT
14.57 GB
VRAM
—
MTP
282.3 W
Power draw
L2 batch seed 2026-10-19: llama.cpp on hw_id=14 running model_id=52 at IQ3_S