Can an RTX 3060 12GB run Qwen3 32B?
Barely
It only loads at IQ2_XXS, which compresses the weights far enough to visibly degrade the model, so we would not call that running it.
In practice
It loads, and we would not call that running it. The only quantization that fits the GeForce RTX 3060 12GB is IQ2_XXS, which compresses the weights far enough that Qwen3 32B loses a visible amount of its ability, so we do not recommend it. At Q4_K_M it needs 20.42 GB against the 11.2 GB available.
Running it properly means a split with system RAM: at Q4_K_M about 34 of 64 layers fit on the card, and the estimate falls to 2.8 tokens per second. A smaller model at a proper quantization is a better answer than a larger one squeezed in, and the next section lists the ones that fit.
The numbers
Qwen3 32B has 32.76 billion parameters across 64 layers. The GeForce RTX 3060 12GB has 12 GB of VRAM at 360 GB/s, of which about 11.2 GB is left for a model once the operating system has taken its share.
| Quantization | Total needed | Fits | Speed |
|---|---|---|---|
| Q8_0 | 34.09 GB | partial | 1.2 t/s |
| Q6_K | 26.86 GB | partial | 1.8 t/s |
| Q5_K_M | 23.63 GB | partial | 2.2 t/s |
| Q4_K_M | 20.42 GB | partial | 2.8 t/s |
| IQ4_XS | 18.26 GB | partial | 3.5 t/s |
| Q3_K_M | 17 GB | partial | 4.2 t/s |
| IQ3_XXS degraded | 13.83 GB | partial | 7.6 t/s |
| Q2_K degraded | 14.91 GB | partial | 5.9 t/s |
| IQ2_XXS degraded | 10.11 GB | yes | 27.4 t/s |
1 of the 9 levels fit entirely. Levels marked degraded are listed for completeness, not as advice. Rows marked partial or experts in RAM run with part of the model in system RAM, at the speed shown.
What to run instead
A smaller model from the same family is the straightforward answer. These fit on an RTX 3060 12GB at a quantization worth using:
- Qwen3 14B, 14.77B at Q5_K_M, 10.96 GB, about 25.2 tokens/s
- Qwen2.5 14B Instruct, 14.77B at Q4_K_M, 9.81 GB, about 28.3 tokens/s
- Qwen2.5 Coder 14B Instruct, 14.77B at Q4_K_M, 9.81 GB, about 28.3 tokens/s
Shortening the context will not rescue this one: at 4k it still wants 19.42 GB against 20.42 GB at 8k, because the weights, not the cache, are what fill the card.
Models of this size are what a second GPU or a large-memory machine is for. The multi-GPU guide shows what 2 cards add, and the Mac and PC comparison shows where unified memory holds a model that no single card can.
Or the hardware that does run it
The cheapest device we track that runs Qwen3 32B at a quantization worth using is the Radeon RX 7900 XT, around $650 , running it at IQ4_XS at about 28.0 tokens per second. Check the current price (affiliate link, indicative price reviewed 2026-09-01). Compare every option.
What else an RTX 3060 12GB runs well
For comparison, these are the other popular models that fit this device at a sound quantization, best pairings first.
| Model | Quantization | Memory | Speed |
|---|---|---|---|
| Gemma 4 12B Instruct | Q6_K | 10.84 GB | 25.4 t/s |
| Qwen3.5 9B | Q8_0 | 9.25 GB | 30.0 t/s |
| Qwen3 14B | Q5_K_M | 10.96 GB | 25.2 t/s |
| Qwen3 8B | Q8_0 | 9.02 GB | 30.8 t/s |
The same model on nearby hardware
Memory decides whether Qwen3 32B loads and bandwidth decides how quickly it answers. Here is what a step down, a card with the same memory and a step up do with it.
| Device | Memory | Bandwidth | Result |
|---|---|---|---|
| GeForce RTX 3070 | 8 GB | 448 GB/s | partial, 2.1 t/s |
| GeForce RTX 4070 | 12 GB | 504 GB/s | does not fit |
| GeForce RTX 5080 | 16 GB | 960 GB/s | does not fit |
Questions people ask
How much VRAM does Qwen3 32B need on RTX 3060 12GB?
At Q4_K_M with an 8k context window it needs 20.42 GB in total: 17.98 GB of weights, 2 GB of KV cache and 0.44 GB of runtime overhead. The GeForce RTX 3060 12GB makes about 11.2 GB available once the operating system has taken its share.
What should I run instead of Qwen3 32B on RTX 3060 12GB?
Qwen3 14B fits at Q5_K_M and runs at about 25.2 tokens per second. Smaller models from the same family are the usual answer.
Keep going
Everything the GeForce RTX 3060 12GB runs, the full breakdown for Qwen3 32B, how the numbers are made in the methodology, or a different pairing in the calculator.