Question 1

How much VRAM does Nemotron-3-Super-120B-A12B-BF16-MTPv2 need?

Accepted Answer

Nemotron-3-Super-120B-A12B-BF16-MTPv2 (2.9B) requires approximately 3336.226620483398 GB of VRAM at full precision (FP16), 1668.1 GB at 8-bit quantization (Q8), and 834.1 GB at 4-bit quantization (Q4). The actual VRAM usage may be slightly higher due to KV cache and runtime overhead.

Question 2

What GPUs can run Nemotron-3-Super-120B-A12B-BF16-MTPv2?

Accepted Answer

We currently have no compatible GPUs indexed for Nemotron-3-Super-120B-A12B-BF16-MTPv2. This may be because the model's VRAM requirements exceed available GPUs in our database, or VRAM data hasn't been computed yet.

Question 3

What is the cheapest GPU to run Nemotron-3-Super-120B-A12B-BF16-MTPv2?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with Nemotron-3-Super-120B-A12B-BF16-MTPv2. Cloud pricing data is updated regularly — check back soon.

Question 4

Can Nemotron-3-Super-120B-A12B-BF16-MTPv2 run on consumer hardware?

Accepted Answer

Nemotron-3-Super-120B-A12B-BF16-MTPv2 requires approximately 3336.226620483398 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 834.1 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so Nemotron-3-Super-120B-A12B-BF16-MTPv2 requires datacenter or cloud GPUs.

Question 5

What quantization options are available for Nemotron-3-Super-120B-A12B-BF16-MTPv2?

Accepted Answer

Nemotron-3-Super-120B-A12B-BF16-MTPv2 supports the following quantization levels: FP16 (full precision): ~3336.226620483398 GB, Q8 (8-bit): ~1668.1 GB, Q6 (6-bit): ~1251.1 GB, Q4 (4-bit): ~834.1 GB. Lower quantization reduces VRAM usage at the cost of some quality loss. Q8 offers a good balance between quality and memory savings, while Q4 provides maximum compression. As a Mixture of Experts (MoE) model with 512 experts (22 active), Nemotron-3-Super-120B-A12B-BF16-MTPv2 has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

Nemotron-3-Super-120B-A12B-BF16-MTPv2

VRAM Requirements

Compatible GPUs

Nemotron-3-Super-120B-A12B-BF16-MTPv2 FAQ

VRAM requirements for Nemotron-3-Super-120B-A12B-BF16-MTPv2 at different quantization levels
Quantization	VRAM Required
FP16	3336.2 GB
Q8	1668.1 GB
Q6	1251.1 GB
Q4	834.1 GB