Qwen3.6-27B · UD-Q3_K_XL
This page aggregates 1 real-world runs of Qwen3.6-27B (UD-Q3_K_XL) on NVIDIA RTX 3090 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.
36.67 tok/s
Decode
Decode speed
39.44 tok/s
Prefill
Prefill speed
— s
TTFT
Time to first token
20.11 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
GitHub
Source platforms
5 days ago
Last verified
Performance
- llama.cpp · Q4_K_M Decode 57.02 · Prefill 1299.01 ·
- llama.cpp · UD-Q3_K_XL (current)Decode 36.67 · Prefill 39.44 ·
- llama.cpp · Q5_K_S Decode 46.57 · Prefill 1201.22 ·
Core figures
- Decode (avg)
- 36.67 tok/s
- Prefill (avg)
- 39.44 tok/s
- TTFT (avg)
- — s
- VRAM (avg)
- 20.11 GB
- MTP acceptance rate
- —
- TTFB
- — GB
- Power draw
- — W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.6-27B
- Quantization
- UD-Q3_K_XL
- Framework
- llama.cpp
- Version
- b9014
- Context length
- 262144 tokens
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 3090
- Nominal VRAM
- 24 GB
- Measured VRAM (avg)
- 20.11 GB
- OS
- Ubuntu 24.04.4 LTS
- Driver
- 580.126.09
- CUDA
- 13.0
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
b9014 · Ubuntu 24.04.4 LTS · CUDA 13.0 · 262144 ctx
36.67 tok/s
Decode
39.44 tok/s
Prefill
— s
TTFT
20.11 GB
VRAM
—
MTP
— W
Power draw
slot print_timing: id 0 | task 217317 | prompt eval time = 101.43 ms / 4 tokens ( 25.36 ms per token, 39.44 tokens per second) eval time = 9518.04 ms / 349 tokens ( 27.27 ms per token, 36.67 tokens per second) total time = 9619.47 ms / 353 tokens slot release: id 0 | task 217317 | stop processing: n_tokens = 898, truncated = 0