# Can I run Qwen3-30B-A3B on an RTX 5090? | What Can I Run AI?

> Yes. Qwen3-30B-A3B at Q4_K_M needs about 22.3 GB; RTX 5090 has 32.0 GB usable. Estimated tokens/sec and context headroom.

Source: https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-5090/

[Home](https://whatcanirunai.vercel.app/) / [Can I run it?](https://whatcanirunai.vercel.app/can-i-run/) / [Qwen3-30B-A3B](https://whatcanirunai.vercel.app/models/qwen3-30b-a3b/) / Qwen3-30B-A3B on an RTX 5090

**Yes.** Qwen3-30B-A3B at Q4\_K\_M needs about 22.3 GB, and the GeForce RTX 5090 has 32.0 GB usable at a \~4k context.

# Can I run Qwen3-30B-A3B on an RTX 5090?

Qwen3-30B-A3B at Q4\_K\_M needs about 22.3 GB, and the GeForce RTX 5090 has 32.0 GB usable. That figure is 32 GB of VRAM on the spec sheet, of which 32.0 GB is usable for weights and cache, at 1792 GB/s memory bandwidth. A Qwen3 Mixture-of-Experts with 30B total weights and about 3B active per token, so it needs 30B-class memory but can decode closer to a small dense model. These are estimates from [our formula](https://whatcanirunai.vercel.app/method/), not measured benchmarks.

Quant that fits

Q4\_K\_M Fits with headroom

Memory needed

22.3 GB at Q4\_K\_M, \~4k context

Usable VRAM

32.0 GB of 32 GB VRAM

Est. tokens/sec

up to 837 estimate

[Qwen3-30B-A3B](https://whatcanirunai.vercel.app/models/qwen3-30b-a3b/) · [GeForce RTX 5090](https://whatcanirunai.vercel.app/hardware/rtx-5090/) · [Everything this machine runs](https://whatcanirunai.vercel.app/what-can-i-run/rtx-5090/) · [Q4 vs Q8 vs FP16](https://whatcanirunai.vercel.app/quantization/)

## Memory at each quant

| Quant    | Need vs 32.0 GB | Verdict       |
| -------- | --------------- | ------------- |
| Q4\_K\_M | 22.3 GB         | Q4Fits        |
| Q8\_0    | 38.3 GB         | Q8Won’t fit   |
| FP16     | 71.2 GB         | FP16Won’t fit |

## Estimated tokens per second

Estimate: **up to 837 tokens/sec** at Q4\_K\_M (upper bound: this is a Mixture-of-Experts, and only the active weights are counted). This is a bandwidth-derived range, not a bench. Efficiency is a 60–85% GPU band from [How we calculate](https://whatcanirunai.vercel.app/method/).

## Context headroom

At Q4\_K\_M, this machine has headroom out to about **64k tokens** under our KV estimate. Longer chats need more memory for the cache. Figures assume an FP16 KV cache.

__Context headroom for Qwen3-30B-A3B on an RTX 5090 at Q4\_K\_M__
| Context | KV cache (FP16) | Total at Q4\_K\_M | Verdict   |
| ------- | --------------- | ----------------- | --------- |
| 4k      | 0.4 GB          | 22.3 GB           | Fits      |
| 8k      | 0.8 GB          | 22.7 GB           | Fits      |
| 16k     | 1.6 GB          | 23.5 GB           | Fits      |
| 32k     | 3.2 GB          | 25.1 GB           | Fits      |
| 64k     | 6.4 GB          | 28.3 GB           | Fits      |
| 128k    | 12.9 GB         | 34.8 GB           | Won’t fit |

From the model’s config. Try other sizes in the [VRAM calculator](https://whatcanirunai.vercel.app/vram-calculator/) or see the [context length guide](https://whatcanirunai.vercel.app/context-length/).

## What to do instead

* Q8\_0 needs 38.3 GB and does not fit here. The [Mac mini M5 Pro 64GB](https://whatcanirunai.vercel.app/hardware/mac-mini-m5-pro-64/) has 48.0 GB usable, enough for Q8.
* See the [best GPU for this model family](https://whatcanirunai.vercel.app/best-gpu-for/qwen3/) and the [Qwen3-30B-A3B model page](https://whatcanirunai.vercel.app/models/qwen3-30b-a3b/).

## Other hardware for Qwen3-30B-A3B

* [Qwen3-30B-A3B on an RTX 3060 12GB](https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-3060-12gb/)Won’t fit
* [Qwen3-30B-A3B on an RTX 3070 Ti](https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-3070-ti/)Won’t fit
* [Qwen3-30B-A3B on an RTX 3080](https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-3080-10gb/)Won’t fit
* [Qwen3-30B-A3B on an RTX 3090](https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-3090/)Tight
* [Qwen3-30B-A3B on an RTX 4060 Ti 16GB](https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-4060-ti-16gb/)Won’t fit
* [Qwen3-30B-A3B on an RTX 4070 Ti Super](https://whatcanirunai.vercel.app/can-i-run/qwen3-30b-a3b/rtx-4070-ti-super/)Won’t fit

## Other models on an RTX 5090

* [Qwen3-32B on an RTX 5090](https://whatcanirunai.vercel.app/can-i-run/qwen3-32b/rtx-5090/)Fits
* [Qwen3.8-27B on an RTX 5090](https://whatcanirunai.vercel.app/can-i-run/qwen3.8-27b/rtx-5090/)Fits
* [Gemma 3 27B on an RTX 5090](https://whatcanirunai.vercel.app/can-i-run/gemma-3-27b-it/rtx-5090/)Fits
* [Gemma 4 31B on an RTX 5090](https://whatcanirunai.vercel.app/can-i-run/gemma-4-31b-it/rtx-5090/)Fits
* [gpt-oss-20b on an RTX 5090](https://whatcanirunai.vercel.app/can-i-run/gpt-oss-20b/rtx-5090/)Fits
* [Mistral Small 24B on an RTX 5090](https://whatcanirunai.vercel.app/can-i-run/mistral-small-3.2-24b-instruct-2506/rtx-5090/)Tight

## Quick answers

Can I run Qwen3-30B-A3B on an RTX 5090?

Yes. Qwen3-30B-A3B at Q4\_K\_M needs about 22.3 GB, and the GeForce RTX 5090 has 32.0 GB usable.

How much VRAM does Qwen3-30B-A3B need?

About 22.3 GB at Q4\_K\_M, 38.3 GB at Q8\_0 and 71.2 GB at FP16, estimated for a \~4k context.

How fast is Qwen3-30B-A3B on an RTX 5090?

Our estimate is up to 837 tokens per second at Q4\_K\_M, as an upper bound for this MoE. It is not a measured benchmark.

What context length can I use?

At Q4\_K\_M, our KV estimate supports up to about 64k tokens before the cache pushes the total over usable memory.

---
Agent guide: https://whatcanirunai.vercel.app/llms.txt · JSON API: https://whatcanirunai.vercel.app/developers/