How much VRAM does Qwen3 235B-A22B Instruct need?

About 149.7 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 155.3 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.

Will Qwen3 235B-A22B Instruct run on a 256GB GPU?

Yes — at Q4_K_M and 4K context it needs about 149.7 GB, which leaves 106.3 GB spare on a Mac M5 Ultra 256G. That is the smallest card in this index that clears it comfortably; 2 of 63 do.

Which quantization of Qwen3 235B-A22B Instruct should I use?

Q4_K_M has the lowest published quality loss (2.2%), and Q4_K_M is the level most people run. All 3 levels in the index are Q4_K_M, Q3_K_M, AWQ INT4. Note this is a mixture-of-experts model — all parameters must be resident even though only a fraction are active per token, so the memory cost follows the total, not the active count.

Qwen3 235B-A22B Instruct

235B-A22B

Alibaba Qwen3

Qwen3 flagship MoE (22B active / 235B total). Q4_K_M ~142GB; rivals DeepSeek-R1 class models.

⬇ 18.7K HF downloads♥ 10 likesQwen/Qwen3-235B-A22B-GGUF· stats from 9/23/2026

Pro GPU

41K

Max Context

Quant Variants

GGUF Q4_K_M

Best Quality

97.8%

Accuracy Retained

Calculate VRAM Hugging Face Compare

Quantization Variants

Per-quant VRAM, quality loss, and inference speed on RTX 4090

Measured = site benchmarks · Estimated = formula · Community = public reports

Format	Level	BPW	VRAM	PPL Loss	Speed	Source	Actions
GGUF	Q4_K_M	4.85	142.0 GB	2.2%	10 tok/s	Estimated	Calc HF
GGUF	Q3_K_M	3.87	115.0 GB	4.8%	12 tok/s	Estimated	Calc HF
AWQ	INT4	4	125.0 GB	3.2%	14 tok/s	Estimated	Calc HF

Similar models

Compare with Qwen3 32B

32B

Qwen3 32B Instruct

Alibaba Qwen3

Consumer GPUPro GPU

16.8 GBmin VRAM97.5%accuracy

Qwen3 dense 32B — successor to Qwen2.5-32B with stronger reasoning and thinking mode.

Qwen3 8B Instruct

Alibaba Qwen3

Consumer GPUMac / Apple Silicon

5.1 GBmin VRAM99.4%accuracy

Latest Qwen3 dense 8B with thinking mode. Strong upgrade from Qwen2.5 7B for local deploy.

14B

Qwen3 14B Instruct

Alibaba Qwen3

Consumer GPUMac / Apple Silicon

9.5 GBmin VRAM98.8%accuracy

Qwen3 14B — best balance of reasoning and VRAM in the 2026 Qwen lineup.

30B-A3B

Qwen3 30B-A3B Instruct

Alibaba Qwen3

Consumer GPUMac / Apple Silicon

17.5 GBmin VRAM98.9%accuracy

Qwen3 MoE with only 3B active params. Q4 ~19GB file; outperforms QwQ-32B on 16GB cards.

Running Qwen3 235B-A22B Instruct locally

At Q4_K_M and 4K of context, Qwen3 235B-A22B Instruct needs about 149.7 GB — 135.3 GB of weights, 0.73 GB of KV cache and a 13.6 GB activation buffer. The smallest card in this index that clears that comfortably is the Mac M5 Ultra 256G at 256 GB, and 2 of the 63 cards here do. These are calculated figures, not measurements: the estimate stops counting a card as comfortable at 88% of its VRAM, which is roughly the room a desktop session needs.

What longer context costs

Going from 4K to 32K adds about 5.15 GB, taking the total to 155.3 GB. Weights do not move with context — only the KV cache does, and it grows linearly, so this is the number to watch when planning for long documents. The model's native window is 40K; holding all of it at Q4_K_M would need about 157 GB.

Which build to download

This index tracks 2 formats for it — GGUF, AWQ — across 3 levels. Q4_K_M carries the lowest published perplexity loss at 2.2%. The fastest level measured here is AWQ INT4 at 14 tok/s on an RTX 4090, batch 1. GGUF runs on llama.cpp and Ollama across NVIDIA, AMD and Apple silicon; AWQ and GPTQ target vLLM on CUDA and ROCm; EXL2 is ExLlamaV2 and CUDA only.

Common questions

How much VRAM does Qwen3 235B-A22B Instruct need?: About 149.7 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 155.3 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.
Will Qwen3 235B-A22B Instruct run on a 256GB GPU?: Yes — at Q4_K_M and 4K context it needs about 149.7 GB, which leaves 106.3 GB spare on a Mac M5 Ultra 256G. That is the smallest card in this index that clears it comfortably; 2 of 63 do.
Which quantization of Qwen3 235B-A22B Instruct should I use?: Q4_K_M has the lowest published quality loss (2.2%), and Q4_K_M is the level most people run. All 3 levels in the index are Q4_K_M, Q3_K_M, AWQ INT4. Note this is a mixture-of-experts model — all parameters must be resident even though only a fraction are active per token, so the memory cost follows the total, not the active count.

Where this model fits

Sized at Q4_K_M with a 4K context window, smallest card first. Comfortable means the estimate uses at most 88% of the memory.

Runs comfortably on: Mac M5 Ultra 256G256GB · +42.3 Mac M5 Ultra 512G512GB · +234.3
Tight but possible: Mac M3 Ultra 192G192GB Mac M2 Ultra 192G192GB
Just misses: Mac M4 Max 128G+53.7GB over Mac M3 Max 128G+53.7GB over
Ships in: GGUF AWQ
Compared here: GGUF vs AWQ
Try it yourself: Size it in the calculator Generate the run command

How to actually run this

Deployment guides for this model and this class of hardware.

This page's figures change when the model or the runtime does.Last updated 2026-09-23 RSS → /feed.xml