Token prices for each VocaLingo tool
Rates per second, minute or 1,000 characters for every tool, with examples. Premium does not change these prices.
Updated November 2, 20266 min read
Each tool shows an estimate before it starts. The rates below are what the server actually charges. Charges are rounded up to a whole token, and tiny operations that cost a fraction of a token add up until they reach a whole token. How charging works in general is explained in How tokens work.
Video Translation
Video Translation is charged per second of the source video, only after the translated video is ready. The minimum is 1 token.
| Provider and option | Per second | Per minute |
|---|---|---|
HeyGen (lip-sync) | 4 | 240 |
HeyGen with Enhanced cloning | 8 | 480 |
ElevenLabs | 4.4 | 264 |
ElevenLabs with Lip Sync | 5.2 | 312 |
HeyGen accepts videos up to 30 minutes. If Lip Sync fails, you get the dubbed video without it and pay the ElevenLabs rate without Lip Sync.
Audio Translation
Audio Translation charges for three steps separately. Recognition and translation run automatically and cost very little. The voiceover is generated only when you tap the play button, and that is where most of the cost is.
Speech recognition
| Option in Settings | Price |
|---|---|
ElevenLabs Scribe (default) | about 0.73 tokens per minute, 44 per hour |
OpenAI Transcribe | about 1.2 tokens per minute |
OpenAI Whisper | 1.2 tokens per minute |
Translation of the recognized text and the Summary tab cost fractions of a token for a typical recording.
Voiceover
The voiceover is charged by the length of the translated text. 1,000 characters is roughly one minute of speech.
| Voice | Price |
|---|---|
ElevenLabs | 10 tokens per 1,000 characters, minimum 1 |
OpenAI | 1.8 tokens for every started 150 words |
Voice cloning (MiniMax), Clone from file | 180 tokens for cloning + 12 per 1,000 characters |
My MiniMax voice | 12 per 1,000 characters |
Voice cloning (Qwen 3), Clone from file | 20 per 1,000 characters, minimum 5 |
My Qwen voice | 10.8 per 1,000 characters |
ElevenLabs Dubbing v2 | 4.4 per second of the source audio (264 per minute) |
Example: a one-minute voice message voiced with an ElevenLabs voice costs about 1 token for recognition and translation plus about 10 for the voiceover. With ElevenLabs Dubbing v2 the same minute costs about 264 tokens.
Voice cloning (Qwen 3) is picked automatically for 10 languages when you have not chosen a voice. For it the app currently shows about half of the real price: count on 20 tokens per 1,000 characters, at least 5 per voiceover.
Text to Speech
Text to Speech is charged by the length of the text, up to 10,000 characters per request.
| Voice | Price |
|---|---|
OpenAI | 1.8 tokens for every started 150 words (up to 150 words: 2 tokens) |
ElevenLabs | 12 per 1,000 characters, minimum 1 |
MiniMax | 12 per 1,000 characters (HD voices), 7.2 (Turbo voices) |
Qwen 3 | 20 per 1,000 characters |
Your own MiniMax voice from My voices | 12 per 1,000 characters |
Your own Qwen voice from My voices | 10.8 per 1,000 characters |
For Qwen 3 voices the app currently shows about half of the real price.
Saving your voice to My Voices
Saving a MiniMax voice costs 180 tokens. They are charged before cloning starts and returned automatically if cloning fails. Saving a Qwen 3 voice costs about 1 token. The voice you record in the onboarding tour is saved for free.
Voice Changer
Voice Changer costs 14.4 tokens per minute of the source recording, counted by its real length, minimum 1. A 30-second clip costs 8 tokens, a 5-minute recording (the maximum) 72. Tokens are charged when the result is ready.
Add Captions
Add Captions charges after the captioned video is ready. The price is the same with or without caption translation.
| Mode | Price | Examples |
|---|---|---|
Mirage (vertical 9:16 video up to 5 minutes) | 15 tokens for every started minute | 40 s: 15, 2 min 10 s: 45 |
Standard | about 6 tokens per minute, at least 15 seconds billed | 30 s: 3, 1 min: 6, 5 min: 29, 10 min: 58 |
Text Transcription and Video to Text
Both tools use the same recognition prices as Audio Translation. With the default ElevenLabs Scribe a transcript costs about 0.73 tokens per minute. Video to Text charges at least 1 token per video: 10 minutes cost 8 tokens, an hour 44. The Essence tab and Export to PDF cost fractions of a token.
Text Translation, photo translation and Text Analysis
Text Translation and Text Analysis: fractions of a token for a typical text. They add up and appear as whole tokens in Usage History.
Listen in Text Translation: 10 or 20 tokens per 1,000 characters, depending on the language.
Translate photo: Nano Banana 2.1 costs 4 or 6 tokens depending on the photo size; GPT Image 2.5 costs 6 for a square image and 5 for a landscape or portrait one.
Tourist Translator
Tourist Translator charges for recognition, translation and voiceover. Recognition costs about 1.3 tokens per minute of speech, so an hour of continuous speech in lecture mode is about 80 tokens. Voiceover adds 10 tokens per 1,000 characters, and translation costs fractions of a token. Recognition and voiceover are charged only when they succeed.
Chat
A chat message costs as much as the model spends on reading your message with its context and writing the answer. The table shows VocaLingo tokens per 1 million model tokens (input / output). A short question and answer usually cost a fraction of a token; long chats with files cost more because the whole context is read again.
| Model | Input / output |
|---|---|
GPT 6 Luna | 12 / 60 |
DeepSeek, MiniMax 2.7 | 36 / 144 |
Gemini 3.8 Flash (Auto uses it by default) | 90 / 450 until 31 December 2026 |
GPT 6 Sol, Claude Sonnet 5.5 | 240 / 1,200 |
Gemini 3.1 Pro | 240 / 1,440 |
Grok 4.7 | 240 / 720 |
Claude Opus 5.5 | 480 / 2,400 |
GPT 6 Astra, Claude Fable 5.1 | 1,200 / 6,000 |
Tools inside the chat are charged on top: a web search costs 0.6 to 1.7 tokens depending on the model, an image from Nano Banana 2.1 costs 4.8 tokens (1K), 7.2 (2K) or 15.6 (4K), and an image from GPT Image 2.5 costs 6 to 7.2. Deep Research has no fixed price: it depends on how many pages it reads. Download Video in chat is free.
AI Calls
AI Calls are charged per minute, in 30-second steps. The price depends on the country and network you call and on whether the AI voices the translation. The exact price for your number is shown in the app before the call.
| Destination | Original sound only | Voiced translation both ways |
|---|---|---|
USA | 6 per minute | 28 per minute |
Germany (mobile) | 12 | 34 |
India (mobile) | 17 | 39 |
Kazakhstan (mobile) | 40 | 62 |
Russia (mobile) | 65 | 88 |
Verifying your own number in My Numbers costs the price of the SMS or call (at least 1 token) plus 10 tokens when you confirm the code.
Frequently asked questions
Is anything cheaper with Premium?
No. Tool prices are the same with and without Premium. Premium gives you a monthly or yearly amount of tokens and lets you turn off the watermark on videos. See What VocaLingo Premium includes.
Why was I charged more than the estimate?
The estimate is calculated in the app before processing, and the charge on the server uses the real length of the translated text or file. The biggest gap is with Qwen 3 voices, where the app shows about half of the price. Each charge is listed in Usage History.
Contact support
Describe the problem and we will reply by email. Add the file name or job ID if you have one.