Question 1

How much VRAM does Nemotron-Cascade-2-30B-A3B need?

Accepted Answer

Nemotron-Cascade-2-30B-A3B (31.6B) requires approximately 227.00058202743529 GB of VRAM at full precision (FP16), 113.5 GB at 8-bit quantization (Q8), and 56.8 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-Cascade-2-30B-A3B?

Accepted Answer

We have 2 compatible GPUs indexed for Nemotron-Cascade-2-30B-A3B, including MI325X, MI350X. Compatibility is determined by whether the GPU's VRAM can fit the model's memory footprint, potentially using quantization to reduce memory usage.

Question 3

What is the cheapest GPU to run Nemotron-Cascade-2-30B-A3B?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with Nemotron-Cascade-2-30B-A3B. Cloud pricing data is updated regularly — check back soon.

Question 4

Can Nemotron-Cascade-2-30B-A3B run on consumer hardware?

Accepted Answer

Nemotron-Cascade-2-30B-A3B requires approximately 227.00058202743529 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 56.8 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so Nemotron-Cascade-2-30B-A3B requires datacenter or cloud GPUs.

Question 5

What quantization options are available for Nemotron-Cascade-2-30B-A3B?

Accepted Answer

Nemotron-Cascade-2-30B-A3B supports the following quantization levels: FP16 (full precision): ~227.00058202743529 GB, Q8 (8-bit): ~113.5 GB, Q6 (6-bit): ~85.1 GB, Q4 (4-bit): ~56.8 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 128 experts (6 active), Nemotron-Cascade-2-30B-A3B has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

Nemotron-Cascade-2-30B-A3B

VRAM Requirements

Compatible GPUs

Nemotron-Cascade-2-30B-A3B FAQ

VRAM requirements for Nemotron-Cascade-2-30B-A3B at different quantization levels
Quantization	VRAM Required
FP16	227 GB
Q8	113.5 GB
Q6	85.1 GB
Q4	56.8 GB

Top 10 compatible GPUs for Nemotron-Cascade-2-30B-A3B, sorted by cheapest price
GPU	VRAM	Best Quant	From	Rent
MI325X	256 GB	Q8	—
MI350X	288 GB	FP16	—