Back to search results
Qwen3.5-0.8B · Q3_K_M
NVIDIA RTX 509032GBvLLM v0.21L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-0.8B (Q3_K_M) on NVIDIA RTX 5090 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
330.24 tok/s
Decode
Decode speed
1,384.87 tok/s
Prefill
Prefill speed
2 s
TTFT
Time to first token
3.24 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
Today
Last verified
Performance
- llama.cpp · Q2_K Decode 378.36 · Prefill 1748.97 ·
- vLLM · Q3_K_M (current)Decode 330.24 · Prefill 1384.87 ·
- vLLM · Q4_K_S Decode 330.77 · Prefill 1336.98 ·
- llama.cpp · IQ2_M Decode 284.64 · Prefill 1115.95 ·
- llama.cpp · Q4_K_M Decode 351.6 · Prefill 1360.11 ·
- vLLM · Q6_K Decode 341.56 · Prefill 1623.94 ·
- Ollama · Q3_K_L Decode 320.91 · Prefill 1586.97 ·
- Ollama · Q8_0 Decode 384.44 · Prefill 1342.7 ·
- Strata · FP16 Decode 337.54 · Prefill 1202.28 ·
- Strata · Q6_K Decode 317.49 · Prefill 1303.51 ·
Core figures
- Decode (avg)
- 330.24 tok/s
- Prefill (avg)
- 1,384.87 tok/s
- TTFT (avg)
- 2 s
- VRAM (avg)
- 3.24 GB
- MTP acceptance rate
- —
- TTFB
- 2.64 GB
- Power draw
- 419.7 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-0.8B
- Quantization
- Q3_K_M
- Framework
- vLLM
- Version
- v0.21
- Context length
- 8192 tokens
- Batch size
- 32
- GPU layers
- 999
- Flash Attention
- On
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 5090
- Nominal VRAM
- 32 GB
- Measured VRAM (avg)
- 3.24 GB
- OS
- Ubuntu 22.04
- Driver
- 565.57
- CUDA
- 12.4
- Power draw
- 419.7 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.21 · Ubuntu 22.04 · CUDA 12.4 · 8192 ctx
330.24 tok/s
Decode
1,384.87 tok/s
Prefill
2 s
TTFT
3.24 GB
VRAM
—
MTP
419.7 W
Power draw
L2 batch seed 2026-10-18: vLLM on hw_id=13 running model_id=51 at Q3_K_M