Model

Qwen3.6 35B A3B

Intelligence rank

#17/216

Speed rank

#18/216

59s

Fastest hardware: NVIDIA GeForce RTX 5090 32GB.

Variations

Intelligence / Disk Size

Variations

Verified model-weight disk size. Higher intelligence and lower disk size are better.

Intelligence Density

Intelligence ÷ Disk Size (GB).

6.75
iq1m · unsloth/Qwen3.6-35B-A3B-GGUF
6.15
iq2xxs
6.04
UD-IQ2_M
5.87
UD-Q2_K_XL
5.40
iq3xxs
5.23
UD-IQ3_S
4.82
UD-Q3_K_S
4.45
q3km
4.40
UD-Q3_K_XL
4.08
iq4xs
3.58
UD-Q4_K_S
3.36
UD-Q4_K_M
3.36
MXFP4_MOE
3.27
q4kxl
3.17
nvfp4 · MTP · k=3 · nvidia/Qwen3.6-35B-A3B-NVFP4
3.13
nvfp4 · unsloth/Qwen3.6-35B-A3B-NVFP4-Fast
2.95
awq-4bit · Qwen3 Next Mtp · k=4
2.83
nvfp4 · unsloth/Qwen3.6-35B-A3B-NVFP4
2.47
nvfp4 · MTP · k=2 · unsloth/Qwen3.6-35B-A3B-NVFP4-Fast
2.25
nvfp4 · MTP · k=3 · unsloth/Qwen3.6-35B-A3B-NVFP4
1.04
Default · MTP · k=1
1.04
Default · Qwen/Qwen3.6-35B-A3B

Hardware

Cost / task speed

Hardware cost / task speed

22 of 22 variations

Model variations

Minimum intelligence

5874

Any intelligence

Each coloured line is that variation’s cost / task-speed Pareto frontier. Global is the frontier across every qualifying variation; every point is still one original measured configuration.