Question 1

How much VRAM does Falcon-H1-34B-Base need?

Accepted Answer

Falcon-H1-34B-Base (33.6B) requires approximately 75.19688358306884 GB of VRAM at full precision (FP16), 37.6 GB at 8-bit quantization (Q8), and 18.8 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 Falcon-H1-34B-Base?

Accepted Answer

We have 10 compatible GPUs indexed for Falcon-H1-34B-Base, including A100 80GB PCIe, A100 80GB, H100 MEGA, H100 SXM, H100 NVL 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 Falcon-H1-34B-Base?

Accepted Answer

The cheapest GPU to run Falcon-H1-34B-Base in the cloud is the A100 80GB PCIe at $1.39/hr. Other affordable options include A100 80GB ($1.39/hr) and H100 MEGA ($1.80/hr). Prices reflect the lowest on-demand rate across cloud providers.

Question 4

Can Falcon-H1-34B-Base run on consumer hardware?

Accepted Answer

Yes, Falcon-H1-34B-Base 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 Falcon-H1-34B-Base?

Accepted Answer

Falcon-H1-34B-Base supports the following quantization levels: FP16 (full precision): ~75.19688358306884 GB, Q8 (8-bit): ~37.6 GB, Q6 (6-bit): ~28.2 GB, Q4 (4-bit): ~18.8 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
A100 80GB PCIe	80 GB	Q8	$1.39/hr	Rent
A100 80GB	80 GB	Q8	$1.39/hr	Rent
H100 MEGA	80 GB	Q8	$1.80/hr	Rent
H100 SXM	80 GB	Q8	$1.80/hr	Rent
H100 NVL	94 GB	FP16	$1.80/hr	Rent
H100 PCIe	80 GB	Q8	$1.80/hr	Rent
RTX Pro 6000 Blackwell	96 GB	FP16	$1.89/hr	Rent
MI300X	192 GB	FP16	$1.99/hr	Rent
GH200	96 GB	FP16	$2.29/hr
H200 SXM	141 GB	FP16	$2.35/hr	Rent

Falcon-H1-34B-Base

VRAM Requirements

Compatible GPUs

Falcon-H1-34B-Base FAQ

VRAM requirements for Falcon-H1-34B-Base at different quantization levels
Quantization	VRAM Required
FP16	75.2 GB
Q8	37.6 GB
Q6	28.2 GB
Q4	18.8 GB