Qwen

Qwen3-Coder 480B

480B-parameter (35B active) MoE coding model with native 262,144-token context, built for agentic software engineering.Apache 2.0Hugging Face

No server in stock right now has the 1152 GB this model needs for inference.

GPU memory it needs

From model cards and vendor documentation. Lower precision needs less memory, with slightly lower quality. Estimates are marked.

What you want to do	Precision	Minimum	Recommended	Setup	Basis
Run the model vLLM recipe minimum for this repo's BF16 checkpoint.	BF16, full	1152 GB	2256 GB	2x 8-GPU H200 nodes (vLLM recipe default, multi-node)	Published
Run the model vLLM recipe minimum for separate repo Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8.	FP8, 8-bit	576 GB	—	—	Published
Run the model vLLM recipe minimum for separate repo nvidia/Qwen3-Coder-480B-A35B-Instruct-NVFP4.	4-bit	288 GB	—	Blackwell GPUs, per vLLM recipe	Published