How much VRAM does Qwen 3 MoE 235B-A22B Instruct need?
Qwen 3 MoE 235B-A22B Instruct needs about 140.1 GB of VRAM at Q4_K_M, or 244.9 GB at Q8_0 and 460.4 GB at FP16. No single GPU in our set holds it even at Q4_K_M; it needs multi-GPU or offload. Weights plus runtime overhead. Budget separately for KV cache, which at 128K is a material share.
Weights-plus-runtime footprint across the six most common quantizations. KV cache is context-dependent and comes on top. At 128K it is gigabytes on its own, before the weights; size it precisely in the GPU Memory Calculator.
This is a mixture-of-experts model: all 235B parameters must sit in memory, but only about 22B are active per token — so it loads like a 235B model and runs closer to a 22B one.
Can your GPU run Qwen 3 MoE 235B-A22B Instruct?
Pick your card — the answer below is computed by the same engine that produces every figure on this page.
H200 141GB cannot hold Qwen 3 MoE 235B-A22B Instruct at Q4_K_M: it needs 140.1 GB against 133.9 GB usable; a heavier quantization, a larger card, or splitting across GPUs are the ways forward; as a mixture-of-experts model all 235B parameters must be resident even though only 22B are active per token.
Weights plus runtime overhead. KV cache is not included — it depends on context length and this model’s published architecture.
| Weights (Q4_K_M) | 132.7 GB |
| Runtime overhead | 7.4 GB |
| Estimated total | 140.1 GB |
| Usable VRAM (95% of 141 GB) | 133.9 GB |
Across every GPU we track
- RTX 3060 12GBNot enough-128.7 GB
- RTX 4060 Ti 16GBNot enough-124.9 GB
- RTX 4070 Ti SUPERNot enough-124.9 GB
- RTX 3090Not enough-117.3 GB
- RTX 4090Not enough-117.3 GB
- RTX 5090Not enough-109.7 GB
- A100 40GBNot enough-102.1 GB
- RTX A6000Not enough-94.5 GB
- L40SNot enough-94.5 GB
- A100 80GBNot enough-64.1 GB
- H100 80GBNot enough-64.1 GB
- H200 141GBNot enough-6.1 GB
Estimates, not guarantees: real usage moves with runtime, driver, batch size and context. Size a specific context window in the GPU Memory Calculator.
How was this number calculated? Every figure here comes from Bitpute’s documented calculation methodology — parameters, precision, quantization, runtime overhead and usable VRAM, each formula written out in full.
Memory by quantization
| Quant | Weights | + runtime overhead |
|---|---|---|
| FP16 | 437.7 GB | 460.4 GB |
| Q8_0 | 232.5 GB | 244.9 GB |
| Q6_K | 179.5 GB | 189.2 GB |
| Q5_K_M | 155.7 GB | 164.2 GB |
| Q4_K_M | 132.7 GB | 140.1 GB |
| Q4_0 | 123.1 GB | 130.0 GB |
Overhead = 0.75 GB + 5% of weights (CUDA context, buffers). Add KV cache on top. Weights dominate at this size, but a 128K window still costs gigabytes.
Which quantization should you actually run?
Qwen 3 MoE 235B-A22B Instruct is a multi-GPU or datacenter model: even Q4_K_M needs about 140.1 GB, beyond any single card. Plan for several 80 GB GPUs or a hosted endpoint; FP16 (460.4 GB) is server-only. As a mixture-of-experts model it must hold all 235B in VRAM but activates only about 22B per token, so it runs much faster than its footprint suggests — memory is the limit here, not speed.
Single-GPU fit at Q4_K_M (140.1 GB + KV)
RTX 3060 12GBRTX 4060 Ti 16GBRTX 4070 Ti SUPERRTX 3090RTX 4090RTX 5090A100 40GBRTX A6000L40SA100 80GBH100 80GBH200 141GB
Common questions
How much VRAM does Qwen 3 MoE 235B-A22B Instruct need?
Qwen 3 MoE 235B-A22B Instruct needs about 140.1 GB of VRAM at Q4_K_M, 244.9 GB at Q8_0, or 460.4 GB at FP16. That is weights plus about 0.75 GB of runtime overhead and 5% of weight size; KV cache is additional and depends on context length.
What GPU can run Qwen 3 MoE 235B-A22B Instruct?
No single GPU in our set fits it at Q4_K_M; it needs multiple GPUs or CPU offload. At FP16 no single GPU in our set is sufficient.
Can Qwen 3 MoE 235B-A22B Instruct run on 24 GB?
Not at Q4_K_M: it needs about 140.1 GB, more than the ~22.8 GB usable on a 24 GB card. Use a larger card, multiple GPUs, or offload.