Question 1

How much VRAM does xLAM-8x22b-r need?

Accepted Answer

xLAM-8x22b-r (140.6B) requires approximately 314.3327041625976 GB of VRAM at full precision (FP16), 157.2 GB at 8-bit quantization (Q8), and 78.6 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 xLAM-8x22b-r?

Accepted Answer

We currently have no compatible GPUs indexed for xLAM-8x22b-r. 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 xLAM-8x22b-r?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with xLAM-8x22b-r. Cloud pricing data is updated regularly — check back soon.

Question 4

Can xLAM-8x22b-r run on consumer hardware?

Accepted Answer

xLAM-8x22b-r requires approximately 314.3327041625976 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 78.6 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so xLAM-8x22b-r requires datacenter or cloud GPUs.

Question 5

What quantization options are available for xLAM-8x22b-r?

Accepted Answer

xLAM-8x22b-r supports the following quantization levels: FP16 (full precision): ~314.3327041625976 GB, Q8 (8-bit): ~157.2 GB, Q6 (6-bit): ~117.9 GB, Q4 (4-bit): ~78.6 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 8 experts (2 active), xLAM-8x22b-r has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

xLAM-8x22b-r

VRAM Requirements

Compatible GPUs

xLAM-8x22b-r FAQ

VRAM requirements for xLAM-8x22b-r at different quantization levels
Quantization	VRAM Required
FP16	314.3 GB
Q8	157.2 GB
Q6	117.9 GB
Q4	78.6 GB