> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.milliseconds.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.milliseconds.ai/_mcp/server.

# Pricing

| Mode                                            | Price per audio hour | Input tokens per audio hour |
| ----------------------------------------------- | -------------------- | --------------------------- |
| With speakers (`diarize=true`)                  | **\$0.12**           | 3,000,000                   |
| Without speakers (`diarize=false`, the default) | **\$0.10**           | 2,500,000                   |

`rakeaudio` bills in input tokens at \$0.04 per million, like every milliseconds.ai model. The same credits pay for `decision-machine-1` and `rakeaudio`. See [Pricing](/reference/pricing).

## How a call is billed

* The API bills the exact audio `duration` of the response, rounded up to a whole token.
* Every successful response carries `x-input-tokens`, the tokens billed for that call.
* A failed call is not billed.

The 19-second example in [Transcribe](/rakeaudio/transcribe) lasts 19.0075 seconds. With speakers it billed 15,840 input tokens, about \$0.0006. Without speakers it billed 13,200.

## Test keys

Test keys use the 125 million monthly test tokens. Those tokens cover about 41 hours of audio with speakers, or 50 hours without. Production keys use purchased credits. See [Authentication](/authentication).