Pay for what you use.
Pay-as-you-go by modality. No subscription, no seats. Every model is metered per token, per character, or per minute.
Models and rates
Addis Voice 2
Lifelike speech in every supported language at one flat rate.
Addis-Aleph-1
Translation is included, no separate endpoint pricing.
Flagship African-language model for chat, generation, and translation.
Addis Scribe
Billed on transcribed text only, audio input is free. About ETB 1.40 per minute of speech.
Accurate speech to text for Amharic and Afan Oromo.
Voice is metered per generated audio minute, the same flat rate for every language. Character limits still protect generation quality, but billing follows the final audio duration.
In beta available now, pricing may change
Addis Live
Native-like speech to speech with under 300ms response. Built for live audio.
Deprecated existing integrations only, use Addis Voice for new work
Addis Voice 1 · AM
Previous-generation Amharic voice. Superseded by Addis Voice.
Addis Voice 1 · OM
Previous-generation Afan Oromo voice. Superseded by Addis Voice.
Worked examples
New to tokens? A token is a small chunk of text the model reads, roughly a short word or a couple of characters. Text models bill per token and voice bills per generated minute, so the examples below show what real tasks come to.
| Task | Detail | Cost |
|---|---|---|
| Narrate a 500-word article | Addis Voice, ~4 minutes of generated audio | $20 |
| Summarize an article | Addis-Aleph, ~400 input + 150 output tokens | $0.0015 |
| Transcribe a 1-minute voice note | Addis Scribe, ~400 output characters (input free) | $0.0089 |
| One minute of live support | Realtime, ~1k input + 1k output tokens | $0.032 |