Can I run Qwen3.5-9B on an RTX 3060 12 GB?

Yes, with a GGUF version

The published weights are too big, but the Q8_0 file from unsloth/Qwen3.5-9B-GGUF (9.53 GB) fits, with room for ≈ 36K tokens of context.

Full analysis of Qwen3.5-9B Everything that runs on RTX 3060 12 GB Get the Q8_0 file on Hugging Face

ollama run hf.co/unsloth/Qwen3.5-9B-GGUF:Q8_0

Files and context on RTX 3060 12 GB

Each file of Qwen3.5-9B (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

File	Size	On RTX 3060 12 GB	Context (16-bit)	Context (8-bit)
As published	19.3 GB	Doesn't fit	—	—
BF16unsloth/Qwen3.5-9B-GGUF	17.9 GB	Doesn't fit	—	—
UD-Q8_K_XLunsloth/Qwen3.5-9B-GGUF	13.0 GB	Doesn't fit	—	—
Q8_0unsloth/Qwen3.5-9B-GGUF	9.53 GB	Tight	36K	73K
UD-Q6_K_XLunsloth/Qwen3.5-9B-GGUF	8.76 GB	Tight	59K	119K
Q6_Kunsloth/Qwen3.5-9B-GGUF	7.46 GB	Fits	98K	196K
UD-Q5_K_XLunsloth/Qwen3.5-9B-GGUF	6.74 GB	Fits	119K	239K
Q5_K_Munsloth/Qwen3.5-9B-GGUF	6.58 GB	Fits	124K	249K
Q5_K_Sunsloth/Qwen3.5-9B-GGUF	6.36 GB	Fits	131K	256K
UD-Q4_K_XLunsloth/Qwen3.5-9B-GGUF	5.97 GB	Fits	143K	256K
Q4_1unsloth/Qwen3.5-9B-GGUF	5.84 GB	Fits	146K	256K
Q4_K_Munsloth/Qwen3.5-9B-GGUF	5.68 GB	Fits	151K	256K
Q4_K_Sunsloth/Qwen3.5-9B-GGUF	5.39 GB	Fits	160K	256K
Q4_0unsloth/Qwen3.5-9B-GGUF	5.38 GB	Fits	160K	256K
IQ4_NLunsloth/Qwen3.5-9B-GGUF	5.37 GB	Fits	160K	256K
IQ4_XSunsloth/Qwen3.5-9B-GGUF	5.17 GB	Fits	166K	256K
UD-Q3_K_XLunsloth/Qwen3.5-9B-GGUF	5.05 GB	Fits	170K	256K
Q3_K_Munsloth/Qwen3.5-9B-GGUF	4.67 GB	Fits	181K	256K
Q3_K_Sunsloth/Qwen3.5-9B-GGUF	4.32 GB	Fits	192K	256K
UD-Q2_K_XLunsloth/Qwen3.5-9B-GGUF	4.12 GB	Fits	197K	256K
UD-IQ3_XXSunsloth/Qwen3.5-9B-GGUF	4.02 GB	Fits	201K	256K
UD-IQ2_Munsloth/Qwen3.5-9B-GGUF	3.65 GB	Fits	212K	256K
UD-IQ2_XXSunsloth/Qwen3.5-9B-GGUF	3.19 GB	Fits	225K	256K

Qwen3.5-9B on other hardware

Popular models that run on RTX 3060 12 GB

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.