Question 1

How much VRAM does Moonlight-16B-A3B need?

Accepted Answer

Moonlight-16B-A3B (16B) requires approximately 35.6736300945282 GB of VRAM at full precision (FP16), 17.8 GB at 8-bit quantization (Q8), and 8.9 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 Moonlight-16B-A3B?

Accepted Answer

We have 10 compatible GPUs indexed for Moonlight-16B-A3B, including A40, L40S, A100 40GB, L40, A100 80GB PCIe and 5 more. 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 Moonlight-16B-A3B?

Accepted Answer

The cheapest GPU to run Moonlight-16B-A3B in the cloud is the A40 at $0.44/hr. Other affordable options include L40S ($0.60/hr) and A100 40GB ($0.93/hr). Prices reflect the lowest on-demand rate across cloud providers.

Question 4

Can Moonlight-16B-A3B run on consumer hardware?

Accepted Answer

Yes, Moonlight-16B-A3B can run on consumer GPUs like the RTX Pro 6000 Blackwell (96 GB VRAM) at full precision (FP16).

Question 5

What quantization options are available for Moonlight-16B-A3B?

Accepted Answer

Moonlight-16B-A3B supports the following quantization levels: FP16 (full precision): ~35.6736300945282 GB, Q8 (8-bit): ~17.8 GB, Q6 (6-bit): ~13.4 GB, Q4 (4-bit): ~8.9 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 64 experts (6 active), Moonlight-16B-A3B has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

GPU	VRAM	Best Quant	From	Rent
A40	48 GB	FP16	$0.44/hr	Rent
L40S	48 GB	FP16	$0.60/hr	Rent
A100 40GB	40 GB	Q8	$0.93/hr	Rent
L40	48 GB	FP16	$0.99/hr	Rent
A100 80GB PCIe	80 GB	FP16	$1.39/hr	Rent
A100 80GB	80 GB	FP16	$1.39/hr	Rent
RTX Pro 6000 Blackwell	96 GB	FP16	$1.89/hr	Rent
MI300X	192 GB	FP16	$1.99/hr	Rent
GH200	96 GB	FP16	$2.29/hr
H100 MEGA	80 GB	FP16	$2.39/hr	Rent

Moonlight-16B-A3B

VRAM Requirements

Compatible GPUs

Moonlight-16B-A3B FAQ

VRAM requirements for Moonlight-16B-A3B at different quantization levels
Quantization	VRAM Required
FP16	35.7 GB
Q8	17.8 GB
Q6	13.4 GB
Q4	8.9 GB