GPU catalogue · Google

Google TPU v5e

The cost-efficient inference TPU.

16 GB GPU memory Accelerator

Request GPU

Google TPU v5e

Specifications

Google TPU v5e: public facts

Good for: Cheap high-volume serving on Google Cloud.

Machine	Google TPU v5e
Memory	16 GB HBM2
Architecture	TPU v5e

Models

Which models fit the Google TPU v5e

language models · image models · video models · embedding models.

Explore

More

GPU catalogue AxForge GPU offerings Model catalogue → On the forum → Pricing →

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms