Use separate input and cache counts. This estimate excludes taxes, cache storage and tool fees. Google API pricing
Where to find the three rates
Open the standard paid tier for Neon in Google Gemini API pricing. Use Input price for uncached input, Context caching price for tokens actually read from the implicit cache, and Output price for billed output including thinking. As checked November 20, the values are $3, $0.15 and $15 per million tokens. The rate fields can be edited if the published prices change or if your billing route uses a different schedule.
Enter each input token once
Some usage reports include cached content in the total prompt token count. For a 500,000-token prompt that reports 400,000 cached tokens, enter 100,000 as uncached input and 400,000 as cached input. Re-entering the full 500,000 in the uncached field would charge the cache-hit portion twice in your estimate. Use zero cached input when no hit is observed. Google currently documents implicit caching for Neon, so repeated text alone does not establish that a hit occurred.
Enter billed output from the usage record. The text you can read may be shorter than the number of tokens billed as output because the rate includes thinking. If you are budgeting a request before running it, try both a typical output count and a higher count to see how sensitive the estimate is to long responses.
Estimate retries and repeated jobs
Calculate each attempt separately if it can have a different cache hit count, prompt size or output length. An application-level rejection can still follow a completed, billable request. For an interrupted or failed call, consult available usage and billing records before assigning a cost. For throughput planning, add the costs of observed attempts and divide by the number of accepted results, then scale to your expected monthly volume.
What the total covers
The calculator multiplies the three token counts by their corresponding rates, divides each by one million and adds the results. The estimate excludes tax, grounding, other tool charges, provider markups and cache storage. Cache storage is relevant to pricing routes or models that support it. Google lists explicit caching as unsupported for Neon in the cited standard tier. Use your billing console to reconcile actual spend.
For complete worked examples and the dated rate table, see the Gemini 4 Neon pricing guide.