Question 1

How much VRAM does Hy3 need?

Accepted Answer

Hy3 (298.8B) requires approximately 663.8545989990234 GB of VRAM at full precision (FP16), 331.9 GB at 8-bit quantization (Q8), and 166 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 Hy3?

Accepted Answer

We currently have no compatible GPUs indexed for Hy3. 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 Hy3?

Accepted Answer

We don't currently have cloud pricing data for GPUs compatible with Hy3. Cloud pricing data is updated regularly — check back soon.

Question 4

Can Hy3 run on consumer hardware?

Accepted Answer

Hy3 requires approximately 663.8545989990234 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 166.0 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so Hy3 requires datacenter or cloud GPUs.

Question 5

What quantization options are available for Hy3?

Accepted Answer

Hy3 supports the following quantization levels: FP16 (full precision): ~663.8545989990234 GB, Q8 (8-bit): ~331.9 GB, Q6 (6-bit): ~248.9 GB, Q4 (4-bit): ~166 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 192 experts (8 active), Hy3 has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

Hy3

VRAM Requirements

Compatible GPUs

Hy3 FAQ

VRAM requirements for Hy3 at different quantization levels
Quantization	VRAM Required
FP16	663.9 GB
Q8	331.9 GB
Q6	248.9 GB
Q4	166 GB