Qwen3.5-0.8B · Q3_K_S
This page aggregates 2 real-world runs of Qwen3.5-0.8B (Q3_K_S) on AMD Radeon RX 9070 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
282.09 tok/s
Decode
Decode speed
1,248.73 tok/s
Prefill
Prefill speed
2.31 s
TTFT
Time to first token
3.19 GB
VRAM
VRAM usage
2
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · Q3_K_S (current)Decode 282.09 · Prefill 1248.73 ·
- llama.cpp · Q4_K_M Decode 250.59 · Prefill 955.15 ·
- Ollama · IQ1_S Decode 353.51 · Prefill 1875.31 ·
- llama.cpp · Q3_K_L Decode 236.42 · Prefill 1273.33 ·
- vLLM · IQ1_S Decode 361.98 · Prefill 1724.43 ·
- vLLM · IQ2_M Decode 325.3 · Prefill 1252.35 ·
Core figures
- Decode (avg)
- 282.09 tok/s
- Prefill (avg)
- 1,248.73 tok/s
- TTFT (avg)
- 2.31 s
- VRAM (avg)
- 3.19 GB
- MTP acceptance rate
- —
- TTFB
- 2.26 GB
- Power draw
- 265.55 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-0.8B
- Quantization
- Q3_K_S
- Framework
- llama.cpp
- Version
- v0.39
- Context length
- 32768 tokens
- Batch size
- 256
- GPU layers
- 33
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- AMD Radeon RX 9070
- Nominal VRAM
- 16 GB
- Measured VRAM (avg)
- 3.19 GB
- OS
- Ubuntu 22.04
- Driver
- amdgpu 6.12
- CUDA
- 12.4
- Power draw
- 212.4 W
Sources & evidence
2 measured records in total, each traceable to its original source
v0.39 · Ubuntu 22.04 · CUDA 12.4 · 32768 ctx
298.35 tok/s
Decode
1,208.27 tok/s
Prefill
2.29 s
TTFT
3.19 GB
VRAM
—
MTP
212.4 W
Power draw
L2 batch seed 2026-10-18: llama.cpp on hw_id=21 running model_id=51 at Q3_K_S
v0.21 · Ubuntu 22.04 · CUDA N/A · 8192 ctx
265.82 tok/s
Decode
1,289.19 tok/s
Prefill
2.33 s
TTFT
3.19 GB
VRAM
—
MTP
318.7 W
Power draw
L2 batch seed 2026-10-13: llama.cpp on hw_id=21 running model_id=51 at Q3_K_S