AWS and Accenture report 40–85% token savings on Bedrock
Start here
Accenture applied six token-spend optimization levers across four client engagements on Amazon Bedrock, and AWS says token spend fell by roughly 40–85 percent where savings were measured. Accenture says results are specific to each client's workload and may vary, and it withheld client names for confidentiality. The levers range from configuration-only steps — prompt caching, which AWS says can cut cached input token costs by up to 90 percent, and batch inference, which is 50 percent cheaper — to architectural changes such as intelligent prompt routing, which AWS says saves up to 30 percent. Two of the four engagements applied several changes at once, so the individual contribution of each lever is not separated out. Accenture frames the work through a Cost per Business Transaction framework that ties token consumption to business outcomes rather than raw token counts.
Generated from reports · updated 1 h ago
Report timeline
Follow the reports to see the event from different sides.
- AWS Machine Learning BlogOptimizing AI economics on Amazon Bedrock with Accenture's sequenced approach
Accenture applied six token-spend optimization levers across four client engagements on Amazon Bedrock, cutting token spend by roughly 40–85 percent where savings were measured. The levers span configuration-only steps like prompt caching (up to 90 percent lower cached input token costs) and batch inference (50 percent cheaper), plus architectural changes such as intelligent prompt routing (up to 30 percent savings). Accenture's Cost per Business Transaction framework ties consumption to business outcomes.
Heat for this event
Not enough continuous observations yet to draw a trend.