Skip to content

Qwen3 Coder 480B A35B Instruct VRAM requirements

How much VRAM Qwen3 Coder 480B A35B Instruct needs
4-bit weights: 271 GBWith 8K context: 274 GBWith 32K context: 280 GB8-bit, 8K context: 478 GB
Parameters
480B
Active per token
35B
Longest context
256K
Released
Jul 2025
License
Apache 2.0
Strongest at
Coding

How much VRAM Qwen3 Coder 480B A35B Instruct needs

Memory Qwen3 Coder 480B A35B Instruct needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit894 GB897 GB903 GB926 GB
8-bit475 GB478 GB484 GB507 GB
6-bit367 GB370 GB375 GB399 GB
5-bit318 GB321 GB327 GB350 GB
4-bit271 GB274 GB280 GB303 GB
3-bit219 GB221 GB227 GB250 GB
2-bit168 GB171 GB176 GB200 GB
1.6-bit92.2 GB95.1 GB101 GB124 GB

One conversation, 16-bit context cache, runtime overhead included.

GPUs and Macs that run Qwen3 Coder 480B A35B Instruct

The best way Qwen3 Coder 480B A35B Instruct runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4060 report
RTX 5060Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 5060 report
RTX 3070Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 3070 report
RTX 3080 10 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 3060 12 GB report
Arc B580Too bigPDF: Qwen3 Coder 480B A35B Instruct on Arc B580 report
RTX 4070Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4070 report
RTX 4070 SuperToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4070 Super report
RTX 5070Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: Qwen3 Coder 480B A35B Instruct on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4080 Super report
RTX 5070 TiToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 5070 Ti report
RTX 5080Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: Qwen3 Coder 480B A35B Instruct on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 3090 report
RTX 4090Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: Qwen3 Coder 480B A35B Instruct on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 5090 report
2× RTX 3090Too bigPDF: Qwen3 Coder 480B A35B Instruct on 2× RTX 3090 report
2× RTX 4090Too bigPDF: Qwen3 Coder 480B A35B Instruct on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 AdaToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX 6000 Ada report
L40SToo bigPDF: Qwen3 Coder 480B A35B Instruct on L40S report
2× RTX 5090Too bigPDF: Qwen3 Coder 480B A35B Instruct on 2× RTX 5090 report
4× RTX 3090Too bigPDF: Qwen3 Coder 480B A35B Instruct on 4× RTX 3090 report
RTX PRO 6000 BlackwellToo bigPDF: Qwen3 Coder 480B A35B Instruct on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Mac mini M4 Pro 64 GB report
M4 Max 128 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on M4 Max 128 GB report
M5 Max 128 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on M5 Max 128 GB report
M3 Ultra 96 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on M3 Ultra 96 GB report
M3 Ultra 256 GB2-bit56K30 tok/s22 sPDF: Qwen3 Coder 480B A35B Instruct on M3 Ultra 256 GB report
M3 Ultra 512 GB5-bit184K18 tok/s22 sPDF: Qwen3 Coder 480B A35B Instruct on M3 Ultra 512 GB report
M5 Ultra 512 GB5-bit184K25 tok/s6.8 sPDF: Qwen3 Coder 480B A35B Instruct on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GBToo bigPDF: Qwen3 Coder 480B A35B Instruct on Ryzen AI Max+ 395 128 GB report
RTX Spark1.6-bit88K26 tok/s7.8 sPDF: Qwen3 Coder 480B A35B Instruct on RTX Spark report
DGX Spark1.6-bit88K24 tok/s7.8 sPDF: Qwen3 Coder 480B A35B Instruct on DGX Spark report
Dual DGX Spark3-bit36K11 tok/s4.5 sPDF: Qwen3 Coder 480B A35B Instruct on Dual DGX Spark report
DGX Station GB3003-bit80K140 tok/s0.6 sPDF: Qwen3 Coder 480B A35B Instruct on DGX Station GB300 report
2× RTX PRO 6000 Blackwell2-bit52K71 tok/s1.1 sPDF: Qwen3 Coder 480B A35B Instruct on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell5-bit176K42 tok/s0.6 sPDF: Qwen3 Coder 480B A35B Instruct on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit256K30 tok/s0.3 sPDF: Qwen3 Coder 480B A35B Instruct on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXMToo bigPDF: Qwen3 Coder 480B A35B Instruct on A100 80 GB SXM report
H100 SXMToo bigPDF: Qwen3 Coder 480B A35B Instruct on H100 SXM report
4× H200 NVL8-bit232K68 tok/s0.6 sPDF: Qwen3 Coder 480B A35B Instruct on 4× H200 NVL report
8× H200 NVL8-bit256K68 tok/s0.3 sPDF: Qwen3 Coder 480B A35B Instruct on 8× H200 NVL report

Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.

How to run it

llama.cpp
llama-server -hf unsloth/Qwen3-Coder-480B-A35B-Instruct-GGUF
vLLM
vllm serve Qwen/Qwen3-Coder-480B-A35B-Instruct
MLX (Mac)
mlx_lm.chat --model mlx-community/Qwen3-Coder-480B-A35B-Instruct-4bit

Questions

How much VRAM does Qwen3 Coder 480B A35B Instruct need?

Qwen3 Coder 480B A35B Instruct needs 271 GB for its weights at 4-bit, 274 GB with an 8K context and 280 GB with 32K. At 8-bit, 478 GB.

What is the minimum VRAM for Qwen3 Coder 480B A35B Instruct?

About 95.1 GB at 1.6-bit with an 8K context, the lowest precision we recommend for a model this size. The smallest machine here that runs it without spilling is the M3 Ultra 256 GB.

How big is Qwen3 Coder 480B A35B Instruct?

480B parameters (35B active per token): 894 GB of weights at 16-bit, 271 GB at 4-bit.

Can Qwen3 Coder 480B A35B Instruct run on an RTX 4090?

Not on one card. The smallest machine here that runs it is the M3 Ultra 256 GB.

More models