MagicTools
LLM VRAM Calculator (all models)

Can You Run Qwen3-235B-A22B? VRAM Requirements

Qwen3-235B-A22B has 235B total parameters (MoE, 22B active per token). Running it at Q4_K_M (GGUF) with 16K context takes roughly 147 GB of memory (134 GB weights + 2.9 GB KV cache + runtime overhead). The minimum viable hardware is about AMD MI300X 192GB. All numbers are planning estimates, verified 2026-07-27.

VRAM by quantization (16K context)

QuantizationWeightsTotal neededMinimum hardware
FP16 / BF16438 GB476 GB7× H100 80GB
FP8219 GB239 GBMac M3 Ultra 512GB
Q8_0 (GGUF)232 GB253 GBMac M3 Ultra 512GB
Q6_K (GGUF)179 GB197 GBMac M3 Ultra 512GB
Q5_K_M (GGUF)155 GB171 GBAMD MI300X 192GB
MXFP4116 GB128 GBH200 141GB
Q4_K_M (GGUF)recommended134 GB147 GBAMD MI300X 192GB
Q3_K_M (GGUF)103 GB114 GBMac 128GB unified
Q2_K (GGUF)76.6 GB85.7 GBMac 128GB unified

How context length changes memory (Q4_K_M (GGUF))

ContextKV cacheTotal needed
4K0.7 GB145 GB
16K2.9 GB147 GB
64K11.8 GB156 GB
128K23.5 GB168 GB

KV cache grows linearly with context; FP8 KV cache halves it again (try it in the calculator below).

Customize the estimate

Back

LLM VRAM Calculator — Can I Run It?

Estimate how much GPU memory (VRAM) you need to run open-weight LLMs like Kimi K3, DeepSeek R1, Qwen3, or Llama locally. Pick a model, quantization, and context length — the calculator adds up weights, KV cache, and runtime overhead, then shows which GPUs or Macs can fit it. Runs entirely in your browser: no upload, no signup.

235B total params · 22B active (MoE) · max 128K context

Recommended balance · ~0.61 bytes/param

Estimated memory needed

168 GB

Model weights134 GB
KV cache (131,072 tokens)23.5 GB
Runtime overhead10.7 GB

Will it fit?

RTX 3060 12GB16× needed
RTX 4060 Ti 16GB12× needed
RTX 3090 / 4090 24GB8× needed
RTX 5090 32GB6× needed
A100 40GB5× needed
A100 / H100 80GB3× needed
H200 141GB2× needed
AMD MI300X 192GB✓ fits
B200 192GB✓ fits
Mac 36GB unified✗ too big
Mac 64GB unified✗ too big
Mac 128GB unified✗ too big
Mac M3 Ultra 512GB✓ fits

Assumes ~92% of device memory is usable. Multi-GPU counts are for tensor/pipeline parallel serving (vLLM, SGLang); Apple Silicon uses unified memory via llama.cpp or MLX.

FAQ

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

About 147 GB at the recommended Q4_K_M (GGUF) quantization with 16K context: 134 GB for weights, 2.9 GB for KV cache, plus runtime overhead. Longer context grows the KV cache.

What is the minimum hardware for Qwen3-235B-A22B?

Roughly AMD MI300X 192GB, assuming 92% of device memory is usable. Lower-end setups can try more aggressive quantization (Q3/Q2) at a noticeable quality cost.

Can I run Qwen3-235B-A22B on a Mac?

Yes. At Q4_K_M (GGUF) with 16K context it needs about 147 GB, which fits the unified memory of a Mac M3 Ultra 512GB via llama.cpp or MLX.

Which quantization should I use for Qwen3-235B-A22B?

Q4_K_M is the recommended balance of size and quality; with headroom, Q6_K or Q8_0 reduce quality loss further.

VRAM requirements for other models