Qwen3.8-27B · IQ3_S
This page aggregates 1 real-world runs of Qwen3.8-27B (IQ3_S) on NVIDIA RTX 5070 Ti with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
64.1 tok/s
Decode
Decode speed
480.7 tok/s
Prefill
Prefill speed
0.21 s
TTFT
Time to first token
11.4 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- Strata · Q5_K_S Decode 49.2 · Prefill 369 ·
- Strata · IQ3_S Decode 82.2 · Prefill 616.5 ·
- Strata · IQ4_XS Decode 61.4 · Prefill 460.5 ·
- llama.cpp · Q5_K_S Decode 38.4 · Prefill 288 ·
- llama.cpp · IQ3_S (current)Decode 64.1 · Prefill 480.7 ·
- llama.cpp · IQ4_XS Decode 47.9 · Prefill 359.3 ·
- llama.cpp · Q4_K_M Decode 42.6 · Prefill 319.5 ·
Core figures
- Decode (avg)
- 64.1 tok/s
- Prefill (avg)
- 480.7 tok/s
- TTFT (avg)
- 0.21 s
- VRAM (avg)
- 11.4 GB
- MTP acceptance rate
- —
- TTFB
- — GB
- Power draw
- — W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.8-27B
- Quantization
- IQ3_S
- Framework
- llama.cpp
- Version
- b4200
- Context length
- 4096 tokens
- Batch size
- 512
- 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 5070 Ti
- Nominal VRAM
- 16 GB
- Measured VRAM (avg)
- 11.4 GB
- OS
- Linux
- Driver
- 570.65
- CUDA
- 12.6
Get started
Original model resources
Base model resources may not match this quantization. Use the verified deployment weights above when available.
Sources & evidence
1 measured records in total, each traceable to its original source
b4200 · Linux · CUDA 12.6 · 4096 ctx
64.1 tok/s
Decode
480.7 tok/s
Prefill
0.21 s
TTFT
11.4 GB
VRAM
—
MTP
— W
Power draw
L0 理论估算（add-1000-records-batch3）