What will this curriculum cost me in AWS charges?

[akash-coded]

I would like to work through the whole curriculum but I am paying for AWS myself. Before I start: roughly what will this cost, and where does the money actually go?

I would rather know now than find out on a bill.

[akash-coded]

Reasonable question, and the honest answer has two halves: inference is cheap, infrastructure left running is not.

Where the money goes

Source	Bills for	Order of magnitude
Bedrock model invocation	Tokens in and out	Cents per exercise, a few dollars per module
Lambda	Invocations	Negligible at learning scale
Knowledge Base storage	Vectors and their store	Small but persistent
OpenSearch Serverless	Existing, not usage	The one that surprises people
AgentCore runtimes / gateways	Existing, not usage	Real if left up

The last two are the whole answer. They bill for existing, so a collection you created on a Tuesday and forgot is still charging on Friday whether you called it or not.

Practical guidance

• Tear down at the end of each module and the total is modest — tens of dollars for the full curriculum.
• Leave Module 10's OpenSearch collection and Module 11's runtime up for a month and that changes completely.
• Set a budget alarm before Module 02. It does not stop spending, but it is the difference between noticing on day two and noticing on day thirty.

The teardown checklist is in cost controls — run it after Modules 10, 11 and 14 especially.

Free while you decide. Modules 00, 01 and 15 need no AWS account at all. Neither do rag_by_hand.py, quality_gate.py, or any of the L.A.B. Simulator labs — every one runs offline on stdlib Python. That is a genuine 15+ hours of work at zero cost if you want to try the material before enabling anything.

Deliberately not quoted here: per-token prices. They change, and a stale number in a discussion thread is worse than none. Bedrock pricing is authoritative.