Question 1

How much VRAM does Hermes-4.3-36B need?

Accepted Answer

Hermes-4.3-36B (36.2B) requires approximately 80.8040153503418 GB of VRAM at full precision (FP16), 40.4 GB at 8-bit quantization (Q8), and 20.2 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 Hermes-4.3-36B?

Accepted Answer

We have 10 compatible GPUs indexed for Hermes-4.3-36B, including RTX Pro 6000 Blackwell, H100 NVL, GH200, H200 SXM, MI300X 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 Hermes-4.3-36B?

Accepted Answer

The cheapest GPU to run Hermes-4.3-36B in the cloud is the RTX Pro 6000 Blackwell at $1.10/hr. Other affordable options include H100 NVL ($1.80/hr) and GH200 ($2.29/hr). Prices reflect the lowest on-demand rate across cloud providers.

Question 4

Can Hermes-4.3-36B run on consumer hardware?

Accepted Answer

Yes, Hermes-4.3-36B 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 Hermes-4.3-36B?

Accepted Answer

Hermes-4.3-36B supports the following quantization levels: FP16 (full precision): ~80.8040153503418 GB, Q8 (8-bit): ~40.4 GB, Q6 (6-bit): ~30.3 GB, Q4 (4-bit): ~20.2 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.

GPU	VRAM	Best Quant	From	Rent
RTX Pro 6000 Blackwell	96 GB	FP16	$1.10/hr
H100 NVL	94 GB	Q8	$1.80/hr	Rent
GH200	96 GB	FP16	$2.29/hr
H200 SXM	141 GB	FP16	$2.35/hr	Rent
MI300X	192 GB	FP16	$2.39/hr	Rent
MI350X	288 GB	FP16	$5.49/hr	Rent
B200	192 GB	FP16	$6.99/hr
B100	192 GB	FP16	—
MI325X	256 GB	FP16	—
Gaudi 3	128 GB	FP16	—

Hermes-4.3-36B

VRAM Requirements

Compatible GPUs

Hermes-4.3-36B FAQ

VRAM requirements for Hermes-4.3-36B at different quantization levels
Quantization	VRAM Required
FP16	80.8 GB
Q8	40.4 GB
Q6	30.3 GB
Q4	20.2 GB