Question

A media company runs a document-summarization service on Amazon Bedrock using a single Model Unit of Provisioned Throughput with a 1-month commitment on Anthropic Claude. Two weeks into the term, a new enterprise customer onboards and steady-state traffic doubles, causing sustained ThrottlingExceptions during business hours. The team needs additional dedicated capacity immediately without breaking their existing commitment or introducing per-request latency variability from shared capacity. Which approach best meets these requirements?

Accepted Answer

Purchase an additional Model Unit of Provisioned Throughput (no-commitment or new term) for the same model to add dedicated capacity alongside the existing commitment.

Answer

Cancel the existing 1-month commitment and repurchase a larger commitment covering both Model Units to consolidate billing.

Answer

Route the overflow traffic to the on-demand (InvokeModel) endpoint for the same model and rely on default account quotas to absorb the surge.

Answer

Request a service quota increase for on-demand tokens-per-minute and remove Provisioned Throughput entirely.