Packaging change·Jul 1 → Nov 1, 2024·Q3 2024·AI & Machine Learning

Huggingface Introduces Enhanced ZeroGPU Pro with 5x Usage Quota and Highest GPU Queue Priority, Alongside New L40S GPU Options

HuggingfacePackagingFeature addedThreshold removedFeature changeLimit increase

What changed

Feature added

Centralized token control and approval added to Enterprise Hub

Threshold removed

Nvidia H100 removed from Spaces Hardware on-demand section (was $10.00/hr Custom tier with 24 vCPU, 250GB RAM, 80GB VRAM). H100 remains available in Inference Endpoints.

Feature change

ZeroGPU Pro feature description updated: Changed from 'Use distributed A100 hardware on your Spaces' to 'Get 5x usage quota and highest GPU queue priority' with separate Spaces Hosting feature for A100 hardware

Limit increase

L40S GPU options added to Spaces Hardware: 1x L40S at $1.80/hr (8 vCPU, 62GB, 48GB VRAM), 4x L40S at $8.30/hr (48 vCPU, 382GB, 192GB VRAM), 8x L40S at $23.50/hr (192 vCPU, 1534GB, 384GB VRAM)

Before and after

Jul 1, 2024 vs Nov 1, 2024

Pulse.

100%

Change 1 of 4 · feature added

Centralized token control and approval added to Enterprise Hub

Sign in to view the full comparison

See the before/after screenshots and every marked-up change.

Q2 2024 · Huggingface Announces Major Price Drop: 4x Nvidia A10G-large Spaces Hardware Reduced by 54% to $5.00/hour Huggingface Reduces GCP NVIDIA A100 Inference Endpoint Prices by 40% to $3.60/hour, Delivering Significant Cost Savings · Q4 2024

More from Huggingface

Full history →