Applied AI / Cost explorer

Compare AI running costs

Compare 30-day token costs, test caching, and check your budget. Rates are editable and illustrative. Every calculation stays in your browser.

Your actual prompt

Paste it here

The colored pieces illustrate text segmentation, not billed model tokens. Editing the prompt seeds a rough estimate of one token per four characters; replace it below with a model-specific count for reliable budgeting.

Your workload

Requests per day

Every call that reaches the model.

Input tokens per request

Starts with a rough character-based estimate. Replace it with your provider’s measured count.

Output tokens each

What the model writes back.

Input, per million

Provider price in dollars.

Output, per million

Enter the output rate for your selected model.

Cached input, per million

Enter your provider’s cached-input rate. Defaults are illustrative, not a live price quote.

Prompt cache hit rate 0%

Share of input tokens served from a cached prefix. A stable system prompt is the possible source of cache hits. Eligibility and actual hit rates depend on the provider.

Every month

Where the cost goes

Uncached input: —
Cached input: —
Output: —

Compare two configurations.

Save the current settings as your baseline. Then change the workload, rates, or cache hit rate above to see the difference.

Adjust current settings ↑

No baseline saved yet.

Saved baseline / 30 days: —
Current configuration / 30 days: —

Save a baseline to compare costs.

The baseline stays in this page only. Reloading clears it.

Does it fit the budget?

30-day token budget / USD

The current estimate will be compared with this budget.

This checks token charges only. Leave room for tools, hosting, retries, taxes, and other fees.

This model includes token charges only. It excludes tools, hosting, retries, taxes, and other provider fees. Use your provider’s tokenizer and current prices before making a spending decision. Nothing you type is sent anywhere.