Independent guide · Checked 16 October 2026

Muse Wonder cost calculator

Turn token assumptions into an API budget. Change input, output, cache share and request count; the breakdown updates in your browser.

No API key · Runs in your browser

Estimate your workload.

Rates checked 16 October 2026: $4 input, $0.20 cached input and $20 output per million tokens; $2.50 per 1,000 search queries.

Set your workload

Total input, including the portion served from cache.
Include billable reasoning tokens in your estimate.
A planning assumption. Actual cache hits are reported by the API.
For example, your daily or monthly total.
An average if only some requests search.

Your Muse Wonder estimate

Total for your chosen period $80.00

$0.08 per request · Standard mode · USD

Uncached input$40.00
Cached input$0.00
Billable output$40.00
Search queries$0.00
Same assumptions on Spark Standard:
$21.00 — a cost comparison, not a capability comparison.

Estimate only. Excludes taxes, currency conversion, reseller fees, other tools and variation in actual usage. No API key or personal data required.

Meta’s official price table ↗. Spark Standard comparison: $1.25 input, $0.15 cached input and $4.25 output per million tokens.

The calculation, step by step

Cached tokens are part of total input. They are removed from full-price input before the discounted rate is applied.

Uncached input = input tokens × (1 − cached share)
Cached input = input tokens × cached share
Cost/request = (uncached input × $4 + cached input × $0.20
+ billable output × $20) / 1,000,000
+ search queries × $0.0025
Period cost = cost/request × requests

Worked example: 50% cached input

For 10,000 input tokens, 2,000 billable output tokens and one search query, 50% caching means 5,000 tokens at each input rate. Charges are $0.020 + $0.001 in input, $0.040 in output and $0.0025 for search. Total: $0.0635 per request, or $63.50 for 1,000 requests.

Use real usage counts when available.

The calculator does not tokenize a prompt, call the model, measure reasoning or guarantee cache hits. Replace assumptions with the provider’s reported usage after testing your app.

Three example workloads

Illustrative workloadInput / output per requestCache assumptionTotal estimate
Short chat · 1,000 requests2,000 / 500 tokens0%$18.00
Document review · 100 requests100,000 / 10,000 tokens50%$41.00
Agent workflow · 10,000 requests10,000 / 2,000 tokens80% + 1 search/request$521.00

What to include

Include instructions, conversation history, documents and tool results when they are sent as input. Budget for billable output, including reasoning where reported, rather than only the answer you display.

What is excluded

Taxes, exchange-rate charges, hosting, reseller fees and other tools are outside this estimate. A monthly subscription is not converted into API credits.

Read the billing guide for line items and the cached-input guide for interpreting cache usage.