AmazonBedrockPricing

Question 1

Amazon ____ access can be given to your team to allow them to create AI-powered applications.

Answer 1

Bedrock Studio

Question 2

Amazon Bedrock provides ____ detection to check if an image was generated by the Titan Generator.

Answer 2

watermark

Question 3

With Amazon Bedrock on-demand use, you are charged on a ____ plan.

Answer 3

pay-as-you-go

Question 4

On-Demand Amazon Bedrock pricing only works with ____ models.

Answer 4

base

Question 5

With Amazon Bedrock on-demand pricing text models, you are charged for every input/output ____ processed.

Answer 5

token

Question 6

With Amazon Bedrock on-demand pricing embedding models, you are charged for every input ____ processed.

Answer 6

token

Question 7

With Amazon Bedrock on-demand pricing image models, you are charged for every image ____.

Answer 7

generated

Question 8

You can get Amazon Bedrock discounts up to 50% by using ____ mode.

Answer 8

batch

Question 9

When Amazon Bedrock batch pricing is used, multiple predictions at a time can be submitted and a ____ output files is sent to ____.

Answer 9

single / S3

Question 10

____ allows you to purchase model units for a certain time (1 month, 6 months, etc)

Answer 10

provisioned throughput Bedrock pricing

Question 11

With ____, you are guaranteed a max number of input/output tokens processed per minute.

Answer 11

provisioned throughput Bedrock pricing

Question 12

Provisioned Throughput Bedrock pricing works with ____ models, but is necessary with ____ and ____ models.

Answer 12

base / fine-tuned and custom

Question 13

What are the three Amazon Bedrock pricing options?

Answer 13

on-deman, batch, provisioned throughput

Question 14

Rank the following model improvement techniques from least to most expensive.

a) instruction-based fine-tuning
b) prompt engineering
c) domain adaptation fune-tuning
d) retrieval augmented generatin (RAG)

Answer 14

prompt engineering, rag, instruction-based fine-tuning, domain adaptation fine-tuning

Question 15

Why is domain adaptation the most expensive technique for model improvement?

Answer 15

model is trained on domain-specific dataset (requires intensive computation)

Question 16

Which of the following model improvement techniques does not require any additional model training or changes?

a) instruction-based fine-tuning
b) prompt engineering
c) domain adaptation fune-tuning
d) retrieval augmented generatin (RAG)

Answer 16

prompt engineering and rag

Question 17

Which of the following model improvement techniques require additional computation?

a) instruction-based fine-tuning
b) prompt engineering
c) domain adaptation fune-tuning
d) retrieval augmented generatin (RAG)

Answer 17

instruction-based fine-tuning and domain adaptation fine-tuning

Question 18

What is the main driver of cost savings with Amazon Bedrock?

Answer 18

the number of input and output tokens

Question 19

True/False: Temperature, TopK and Top P parameters can affect your Bedrock pricing.

Answer 19

False: no impact on pricing

Question 20

Normally a ____ model will be cheaper, but varies based on providers.

Answer 20

smaller

Question 21

____ Bedrock pricing is great for unpredictable workloads and has no long-term commitment.

Answer 21

On-Demand

Question 22

____ Bedrock pricing is not a cost-saving measure, but is a way to “reserve” capacity.

Answer 22

Provisioned Throughput