Amazon Nova Pro vs Claude Sonnet 4.5

Two models most teams evaluate side by side inside Bedrock.

Prices verified 4 Aug 2026
Input / 1M$0.800
Output / 1M$3.20
Cached input / 1M
Batch50% off
Context300K tokens
Blended (3:1)$1.40
Input / 1M$3.00
Output / 1M$15.00
Cached input / 1M$0.300
Batch50% off
Context200K tokens
Blended (3:1)$6.00

Monthly cost by workload shape

The same volume costs very different amounts depending on the shape of the traffic. No caching or batch applied.

WorkloadAmazon Nova ProClaude Sonnet 4.5Difference
Customer support chatbot
50K/day · 800 in / 300 out
$2,435.20$10.5KAmazon Nova Prosaves $8,066.60Model it →
RAG / search answers
20K/day · 3000 in / 400 out
$2,240.38$9,132.00Amazon Nova Prosaves $6,891.62Model it →
Agent / tool-use loop
8.0K/day · 2500 in / 600 out
$954.60$4,018.08Amazon Nova Prosaves $3,063.48Model it →
Batch summarization
100K/day · 5000 in / 500 out
$17.0K$68.5KAmazon Nova Prosaves $51.4KModel it →
Code assistant
15K/day · 2000 in / 800 out
$1,899.46$8,218.80Amazon Nova Prosaves $6,319.34Model it →

Frequently asked questions

Which is cheaper, Amazon Nova Pro or Claude Sonnet 4.5?

At a typical 3:1 input-to-output ratio, Amazon Nova Pro is cheaper: $1.40 blended per 1M tokens against $6.00. That holds at every input:output ratio — one model is cheaper on both rates.

What is the price difference between Amazon Nova Pro and Claude Sonnet 4.5?

Amazon Nova Pro: $0.800 input, $3.20 output per 1M tokens. Claude Sonnet 4.5: $3.00 input, $15.00 output. That is 3.8x on input and 4.7x on output.

Does prompt caching change the answer?

Amazon Nova Pro has no prompt caching and Claude Sonnet 4.5 caches at $0.300/1M. For input-heavy workloads with repeated context, caching can matter more than the base rate difference — model both in the calculator rather than comparing rate cards.

This page compares published prices only. It makes no claim about which model performs better on any task — that depends entirely on your evaluations.

Rate cards decide nothing on their own.

Most teams run several models at once. Finitizer TokenOps shows what each one actually costs you in production, by team and feature, so a routing decision is made on data rather than on a price page.