Calculator
Prompt cache savings calculator
Prompt cache savings calculator is built for teams with repeated prefixes, long system prompts, retrieval context, or agent memory. Use it to decide whether cache reads offset cache-write premiums enough to reduce monthly spend. Keep the workload assumptions consistent across options, then inspect the cited prices and last-checked dates before committing budget.
Open cache calculator - Find cache break-even reuse →
The decision this page helps you make
Quantify prompt caching savings from your cache hit rate and reused prefix, including the break-even reuse count where caching pays off.
The practical question is whether cache reads offset cache-write premiums enough to reduce monthly spend. Use the same workload assumptions for every option so the comparison reflects billing differences instead of different inputs.
Start with these inputs
- Prefix: Reusable tokens per request.
- Hit rate: Share of requests reading from cache.
- Savings: Break-even reuse and adjusted monthly cost.
What the result includes
| Area | What ByteCosts shows |
|---|---|
| Prefix | Reusable tokens per request |
| Hit rate | Share of requests reading from cache |
| Savings | Break-even reuse and adjusted monthly cost |
How to use the result
- Run a realistic base case and a heavier-usage case before choosing a provider or plan.
- Compare alternatives with identical traffic, token, seat, runtime, and retry assumptions.
- Open the cited provider source before a purchase or production billing decision.
Formula
effectiveInputCost = cachedInputTokensM * cachedInputPricePerMTok + uncachedInputTokensM * inputPricePerMTok + cacheWriteTokensM * cacheWritePricePerMTok.
Assumptions
- Cacheable prefixes must remain stable enough for provider cache matching.
- Read and write rates use source-backed fields where available and estimates only where clearly labeled.
- A low hit rate or tiny prefix can make caching more expensive than no caching.
- Output tokens are still billed normally.
Example scenario
Model a large shared system prompt with a high hit rate. Compare the no-cache input cost against cache write plus cache read cost to find the reuse break-even point.
How to read the example
| Step | Example input | What to inspect |
|---|---|---|
| No cache | Prefix billed at normal input rate | Baseline monthly cost |
| Cache write | One-time premium for the reusable prefix | Up-front cost |
| Cache reads | Repeated prefix billed at read rate | Savings after reuse break-even |
Interpretation guide
- Caching is strongest for large shared prefixes that are reused many times.
- Watch write-tier choices because longer TTLs cost more to populate.
- Verify savings against invoice data after deployment.
Limitations
Prompt cache savings calculator is a planning tool, not a billing guarantee. It uses the visible assumptions and committed source-backed data available at the page's last update.
Check the cited provider page and your own production logs before signing a contract, changing price, or committing infrastructure spend.
Frequently asked questions
What should I enter first in Prompt cache savings calculator?
Start with prefix: reusable tokens per request. Add optional adjustments only after the base case is understandable.
Is the result a guaranteed invoice forecast?
No. It is a planning estimate based on the visible workload assumptions and source-backed public prices. Taxes, negotiated discounts, undocumented limits, and production behavior can change the final invoice.
Where do the prices and assumptions come from?
ByteCosts keeps provider source links, confidence information, and last-checked dates attached to pricing records. User-entered workload assumptions remain separate from published vendor facts.
Prompt cache savings calculator. ByteCosts. https://bytecosts.com/tools/cache-savings/