Question 1

How much VRAM does Llama-3.2-90B-Vision-Instruct need?

Accepted Answer

Llama-3.2-90B-Vision-Instruct (88.6B) requires approximately 198.02158025577663 GB of VRAM at full precision (FP16), 99 GB at 8-bit quantization (Q8), and 49.5 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 Llama-3.2-90B-Vision-Instruct?

Accepted Answer

We have 2 compatible GPUs indexed for Llama-3.2-90B-Vision-Instruct, 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 Llama-3.2-90B-Vision-Instruct?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with Llama-3.2-90B-Vision-Instruct. Cloud pricing data is updated regularly — check back soon.

Question 4

Can Llama-3.2-90B-Vision-Instruct run on consumer hardware?

Accepted Answer

Llama-3.2-90B-Vision-Instruct requires approximately 198.02158025577663 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 49.5 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so Llama-3.2-90B-Vision-Instruct requires datacenter or cloud GPUs.

Question 5

What quantization options are available for Llama-3.2-90B-Vision-Instruct?

Accepted Answer

Llama-3.2-90B-Vision-Instruct supports the following quantization levels: FP16 (full precision): ~198.02158025577663 GB, Q8 (8-bit): ~99 GB, Q6 (6-bit): ~74.3 GB, Q4 (4-bit): ~49.5 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.

Llama-3.2-90B-Vision-Instruct

VRAM Requirements

Compatible GPUs

Llama-3.2-90B-Vision-Instruct FAQ

VRAM requirements for Llama-3.2-90B-Vision-Instruct at different quantization levels
Quantization	VRAM Required
FP16	198 GB
Q8	99 GB
Q6	74.3 GB
Q4	49.5 GB

Top 10 compatible GPUs for Llama-3.2-90B-Vision-Instruct, sorted by cheapest price
GPU	VRAM	Best Quant	From	Rent
MI325X	256 GB	FP16	—
MI350X	288 GB	FP16	—