Question 1

How much VRAM does sage-ft-mixtral-8x7b need?

Accepted Answer

sage-ft-mixtral-8x7b (46.7B) requires approximately 104.38887634277343 GB of VRAM at full precision (FP16), 52.2 GB at 8-bit quantization (Q8), and 26.1 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 sage-ft-mixtral-8x7b?

Accepted Answer

We have 9 compatible GPUs indexed for sage-ft-mixtral-8x7b, including MI300X, H200 SXM, B200, B100, MI325X and 4 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 sage-ft-mixtral-8x7b?

Accepted Answer

The cheapest GPU to run sage-ft-mixtral-8x7b in the cloud is the MI300X at $1.99/hr. Other affordable options include H200 SXM ($2.35/hr) and B200 ($5.49/hr). Prices reflect the lowest on-demand rate across cloud providers.

Question 4

Can sage-ft-mixtral-8x7b run on consumer hardware?

Accepted Answer

sage-ft-mixtral-8x7b requires approximately 104.38887634277343 GB of VRAM at FP16, and even with 4-bit quantization (Q4) it needs about 26.1 GB. This exceeds the 24 GB available on the highest-end consumer GPU (RTX 4090), so sage-ft-mixtral-8x7b requires datacenter or cloud GPUs.

Question 5

What quantization options are available for sage-ft-mixtral-8x7b?

Accepted Answer

sage-ft-mixtral-8x7b supports the following quantization levels: FP16 (full precision): ~104.38887634277343 GB, Q8 (8-bit): ~52.2 GB, Q6 (6-bit): ~39.1 GB, Q4 (4-bit): ~26.1 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), sage-ft-mixtral-8x7b has a larger total weight size but only activates a subset of parameters per token, which can affect effective memory usage during inference.

GPU	VRAM	Best Quant	From	Rent
MI300X	192 GB	FP16	$1.99/hr	Rent
H200 SXM	141 GB	FP16	$2.35/hr	Rent
B200	192 GB	FP16	$5.49/hr	Rent
B100	192 GB	FP16	—
MI325X	256 GB	FP16	—
MI350X	288 GB	FP16	—
Gaudi 3	128 GB	FP16	—
DGX Spark (GB10)	128 GB	FP16	—
M3 Ultra (192 GB)	144 GB	FP16	—

sage-ft-mixtral-8x7b

VRAM Requirements

Compatible GPUs

sage-ft-mixtral-8x7b FAQ

VRAM requirements for sage-ft-mixtral-8x7b at different quantization levels
Quantization	VRAM Required
FP16	104.4 GB
Q8	52.2 GB
Q6	39.1 GB
Q4	26.1 GB