Back to search results
Qwen3.5-4B · IQ2_XS
AMD Radeon RX 907016GBllama.cpp v0.53L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-4B (IQ2_XS) on AMD Radeon RX 9070 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
204.13 tok/s
Decode
Decode speed
1,113.36 tok/s
Prefill
Prefill speed
1.97 s
TTFT
Time to first token
6.13 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · IQ2_XS (current)Decode 204.13 · Prefill 1113.36 ·
- llama.cpp · Q3_K_L Decode 162.78 · Prefill 836.11 ·
- vLLM · Q3_K_S Decode 194.22 · Prefill 736.68 ·
- vLLM · Q4_K_M Decode 137.14 · Prefill 684.52 ·
- llama.cpp · IQ4_XS Decode 127.63 · Prefill 638.8 ·
- vLLM · IQ1_S Decode 172.53 · Prefill 556.88 ·
- vLLM · IQ3_S Decode 146.59 · Prefill 752.13 ·
Core figures
- Decode (avg)
- 204.13 tok/s
- Prefill (avg)
- 1,113.36 tok/s
- TTFT (avg)
- 1.97 s
- VRAM (avg)
- 6.13 GB
- MTP acceptance rate
- —
- TTFB
- 1.63 GB
- Power draw
- 249.9 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-4B
- Quantization
- IQ2_XS
- Framework
- llama.cpp
- Version
- v0.53
- Context length
- 4096 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
- AMD Radeon RX 9070
- Nominal VRAM
- 16 GB
- Measured VRAM (avg)
- 6.13 GB
- OS
- Ubuntu 24.04
- Driver
- amdgpu-pro 24.20
- CUDA
- 12.6
- Power draw
- 249.9 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.53 · Ubuntu 24.04 · CUDA 12.6 · 4096 ctx
204.13 tok/s
Decode
1,113.36 tok/s
Prefill
1.97 s
TTFT
6.13 GB
VRAM
—
MTP
249.9 W
Power draw
L2 batch seed 2026-10-18: llama.cpp on hw_id=21 running model_id=48 at IQ2_XS