About · Checked Oct 27, 2026

Methodology: how these prices are checked

Prices on VoiceAICost are list prices read from the vendor's own page on Oct 27, 2026. Each figure keeps its source URL and checked date. If a field was not on the page, it is shown as Not published, with the reason, and it is never filled in with an estimate.

Last verified · How prices are checked

How a figure gets onto a page

  1. Read the vendor pricing or docs page.
  2. Store the value, unit, source URL, checked date and status in the price dataset.
  3. Pages and the calculator read that record by id. The same id produces the same number everywhere.

How often

A date changes only when the figure is re-read. The sitemap lastmod, the Last verified line and dateModified move together. Oct 27, 2026 is the first recorded check.

Calculator math

One function prices Retell, Vapi, Bland, ElevenLabs Agents, GoHighLevel and Deepgram. The default scenario is 10 concurrent calls, 3-minute calls, one US number, 30 days. Vapi's token estimate follows 150 output tokens/min; 5 requests per turn; 50% cache-hit assumption; input tokens depend on your prompt. Where a vendor bills the LLM by token and publishes no per-minute price, the stand-in is $0.0128/min, labeled assumption.

Test calls

No billed test calls have been run. The test-call sections say so. A future row will be an invoice amount with a date, not a guess.

Corrections

If a number disagrees with the vendor page, the fix is to re-read that page, update the dataset, and move the checked date. See also the changelog.

Sources

  1. retellai.com ·
  2. docs.vapi.ai ·