Question 1

How much VRAM does Qwen1.5-MoE-A2.7B need?

Accepted Answer

Qwen1.5-MoE-A2.7B (14.3B) requires approximately 30.141581726074218 GB of VRAM at full precision (FP16), 15.1 GB at 8-bit quantization (Q8), and 7.5 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 Qwen1.5-MoE-A2.7B?

Accepted Answer

We have 10 compatible GPUs indexed for Qwen1.5-MoE-A2.7B, including A40, A100 40GB, RTX A6000, A100 80GB PCIe, A100 80GB 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 Qwen1.5-MoE-A2.7B?

Accepted Answer

The cheapest GPU to run Qwen1.5-MoE-A2.7B in the cloud is the A40 at $0.44/hr. Other affordable options include A100 40GB ($0.93/hr) and RTX A6000 ($1.09/hr). Prices reflect the lowest on-demand rate across cloud providers.

Question 4

Can Qwen1.5-MoE-A2.7B run on consumer hardware?

Accepted Answer

Yes, Qwen1.5-MoE-A2.7B can run on consumer GPUs like the RTX A6000 (48 GB VRAM) at full precision (FP16).

Question 5

What quantization options are available for Qwen1.5-MoE-A2.7B?

Accepted Answer

Qwen1.5-MoE-A2.7B supports the following quantization levels: FP16 (full precision): ~30.141581726074218 GB, Q8 (8-bit): ~15.1 GB, Q6 (6-bit): ~11.3 GB, Q4 (4-bit): ~7.5 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 60 experts (4 active), Qwen1.5-MoE-A2.7B 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
A100 40GB	40 GB	FP16	$0.93/hr	Rent
RTX A6000	48 GB	FP16	$1.09/hr
A100 80GB PCIe	80 GB	FP16	$1.39/hr	Rent
A100 80GB	80 GB	FP16	$1.39/hr	Rent
H100 MEGA	80 GB	FP16	$1.80/hr	Rent
H100 SXM	80 GB	FP16	$1.80/hr	Rent
H100 NVL	94 GB	FP16	$1.80/hr	Rent
H100 PCIe	80 GB	FP16	$1.80/hr	Rent
GH200	96 GB	FP16	$2.29/hr

Qwen1.5-MoE-A2.7B

VRAM Requirements

Compatible GPUs

Qwen1.5-MoE-A2.7B FAQ

VRAM requirements for Qwen1.5-MoE-A2.7B at different quantization levels
Quantization	VRAM Required
FP16	30.1 GB
Q8	15.1 GB
Q6	11.3 GB
Q4	7.5 GB