Packaging change·Jul 6 → Oct 3, 2025·Q3 2025·AI & Machine Learning

Huggingface Introduces Cost-Effective AWS NVIDIA H100 Inference Endpoints at $4.50/hour, Reducing Prices from GCP's $10.00/hour

HuggingfacePackagingThreshold added

What changed

Threshold added

AWS NVIDIA H100 added to Inference Endpoints: 1x 80GB at $4.50/hour (previously only GCP H100 at $10.00/hour)

Threshold added

Nvidia H100 GPU option added to Spaces Hardware: 23 vCPU, 240 GB memory, 80 GB VRAM at $4.50/hour

Threshold added

8x Nvidia H100 GPU option added to Spaces Hardware: 184 vCPU, 1920 GB memory, 640 GB VRAM at $36.00/hour

Threshold added

NVIDIA B200 GPU added to Inference Endpoints with multiple configurations: 1x 256GB at $9.25/hour, 2x 512GB at $18.50/hour, 4x 1024GB at $37.00/hour, 8x 2048GB at $74.00/hour

Before and after

Jul 6, 2025 vs Oct 3, 2025

Pulse.

100%

Change 1 of 4 · threshold added

AWS NVIDIA H100 added to Inference Endpoints: 1x 80GB at $4.50/hour (previously only GCP H100 at $10.00/hour)

Sign in to view the full comparison

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

Q2 2025 · Huggingface Introduces New Team Plan and Upgrades PRO Features Amid Significant Price Decrease for A100 Hardware Huggingface Discontinues Key Hardware Options and Persistent Storage Tiers, Impacting User Access and Pricing · Q4 2025

More from Huggingface

Full history →