Workload cost planner

Plan the cost of an AI workload

Describe a workload with structured planning inputs. Pro customers can calculate low, base, and high token-cost estimates across supported Amazon Bedrock models without sharing prompt or response content.

Pro planning tool

Workload assumptions

No prompt or response content is requested or analyzed.

Pro results

What kind of work?

Choose a task after calculator options load.

Request complexity

Choose how demanding a typical request is; Model Optimizer does not inspect your prompt.

Monthly calls

Monthly growth or decline (%)

Planning horizon

AWS region

Pricing mode

Current model Optional baseline

Compare against a callable model with complete pricing in this region and mode. It never changes candidate eligibility or fitness.

Models included Governed test candidates No model selection is needed. Pro results show the current candidate set from the capability catalog, including candidates with unavailable pricing, and you can choose a priced candidate to inspect.

Token-use assumption

Model Optimizer estimate My averages

Uses a low-to-high token range for your selections. No prompt is analyzed.

Loading calculator options…

What you’re comparing

We apply the same estimated token use for your selected task and complexity to each model’s current Bedrock input and output prices. This produces comparable cost ranges—not a quality ranking or model recommendation.

Evidence and freshness

Comparison sources

View evidence and version details

Planning estimate—not a recommendation or guarantee. Test candidates with your own workload.

Model estimate in view

—

Month 1 estimate

Planning-horizon estimate

Share this planning summary

Copies the visible workload, ranges, evidence, freshness, and version context—never prompt or response content, customer, session, or trace identifiers, raw hashes, or internal model identifiers.

Monthly projection

Each month applies the entered growth or decline rate to call volume.

View exact low, base, and high values by month

Month	Calls	Low	Base	High	Current base