Price reduction for GPT models on Amazon Bedrock

What changed

Source summary

AWS cut Bedrock pricing for OpenAI's GPT-5.6 family effective July 30: an 80% price cut for Luna (now $0.20 per million input tokens and $1.20 per million output tokens) and a 20% cut for Terra, applied automatically with no customer action required.

Why it matters to your work

Practical interpretation

If your business runs OpenAI models via AWS Bedrock, costs just dropped automatically — Luna is now up to 80% cheaper — worth checking your AWS bill or reconsidering which model tier you use for cost-sensitive workloads.

When this could help

Relevant now if you use or are evaluating GPT-5.6 Luna or Terra on Bedrock — Luna is now described as one of the most affordable frontier-class models available on the platform.

Amazon Bedrock reduced on-demand inference prices for GPT-5.6 models effective July 30: GPT-5.6 Luna dropped 80% (now $0.20/M input tokens, $1.20/M output tokens), and GPT-5.6 Terra dropped 20%. No customer action needed — reductions apply automatically.

Build it into your routine

SCALA suggested workflow

1. Bedrock model cost review

   If you use GPT-5.6 on Bedrock, check your latest invoice to confirm the reduced rates are reflected and re-evaluate whether Luna's new pricing changes your model selection for cost-sensitive workloads.

What to check

• Terra's new absolute prices are not stated in the source — only the 20% reduction percentage. Regions are not specified. Old prices are not stated, so the only way to verify is against AWS pricing pages or your own invoices. Source ↗