Agentic cost calculator

Agentic workloads re-send a large prompt prefix on every step, so prompt caching dominates the bill in a way a flat price-per-token comparison never shows. Set your real numbers and see the difference.

ModelRegionInput tokens per runOutput tokens per runRuns per dayCache hit rate60%share of input tokens served from a cached prefixBatch share0%share of runs that can tolerate async batch turnaroundCompare against

Per run

$0.00573

Per day

$11.46

Per month

$343.80

Where the money goes

you pay 78%saved by cachesaved by batch

Baseline / month: $441.00; no cache, no batch
Saved by caching: −$97.20
Saved by batch: −$0.00

Batch inference and prompt caching do not combine on Bedrock: batched runs are costed at full batch rates with no cache discount, which is why raising batch share can reduce cache savings.

Measure a prompt

paste what you actually send — the count updates as you type

Every LLM calculator asks for a token count.
This one measures it.

This is an estimate, not a tokenizer. A real BPE vocabulary is 1–2 MB of WebAssembly — roughly ten times this site’s entire JavaScript budget — so the count is computed from a segment-aware heuristic rather than the model’s own vocabulary. It prices words, numbers, punctuation and CJK runs separately instead of assuming four characters per token, which is why the range moves with the kind of content you paste. For an exact figure, use your SDK’s token-counting endpoint.