May 05, 2025 — Snowflake Cortex Provisioned Throughput — General availability

With Provisioned Throughput, a new capability in Snowflake Cortex, you can reserve throughput for managed inference.

Use Provisioned Throughput for the following tasks:

  • Reserve throughput for specific time periods using provisioned throughput units (PTUs).

  • Allocate capacity for supported models.

  • Scale throughput based on workload requirements with minimum and incremental configurations.

Language: English