Qwen3-Coder 30B-A3B-Instruct

Q4_K_XL·30B params·GGUF
checkpoint: unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:UD-Q4_K_XL

All runs (5)

HardwareBackendShapeConc.Gen tok/sTTFTTPOT (ms)Out tokTotalVRAM Δ
GeForce RTX 3090 · 24 GiBllama.cpp 59778f0 (cuda)codegen1
152.7
165ms6.48115.28s0.000 GiB
GeForce RTX 3090 · 24 GiBllama.cpp 59778f0 (cuda)chat1
146.2
72ms6.099635ms0.000 GiB
GeForce RTX 3090 · 24 GiBllama.cpp 59778f0 (cuda)agent1
141.1
161ms6.52571.80s0.000 GiB
GeForce RTX 3090 · 24 GiBllama.cpp 59778f0 (cuda)rag1
117.1
220ms6.291975ms0.000 GiB
GeForce RTX 3090 · 24 GiBllama.cpp 59778f0 (cuda)agent4
50.5
1.94s13.82575.60s0.000 GiB

Environment

GeForce RTX 3090 · 24 GiB
cpuAMD EPYC 7302P 16-Core Processor
gpuNVIDIA GeForce RTX 3090
archNVIDIA
vram24 GiB (system 64.0 GiB)
power200 W / 450 W max(44% cap)
backendllama.cpp 59778f0 (cuda)
serverlemonade unknown
osUbuntu 24.04 LTS
kernel6.17.13-7-pve
driver590.48.01
python3.12.3
containerizedtrue
runs/cell5
warmups2
endpoint/v1/chat/completions
streamingtrue