Pricing experiment
One offer: provider cost plus a tiny margin.
TalkSpanish.ai is testing an OpenRouter-style model for language learning: charge the underlying AI model and voice cost, then add a clear platform margin for hosting, product work, and support.
Transparent formula
Model input tokens$0.15 / 1M
Model output tokens$0.60 / 1M
Voice characters$0.05 / 1K
Platform margin3%
These defaults are environment-configurable, so the live product can update when OpenAI or ElevenLabs changes provider rates.
Included in every turn
- AI tutor reply through gpt-4o-mini
- Voice playback through eleven_flash_v2_5
- English translation toggle
- Usage receipt after each turn
- MongoDB session history when configured