Supported Models
AI Transcription Notepad uses Google Gemini models via OpenRouter for audio transcription. OpenRouter provides an OpenAI-compatible API with per-key cost tracking and low latency.
Available Models
| Model | Tier | Notes |
|---|---|---|
google/gemini-3-flash-preview | Standard | Default — fast and cost-effective |
google/gemini-3-pro-preview | Premium | Higher quality, used as fallback |
Gemini 3 Flash is currently in preview and is the recommended model for most transcription tasks. Gemini 3 Pro serves as the automatic fallback if Flash is unavailable.
Choosing a Model
Gemini 3 Flash is recommended for most users. It provides an excellent balance of speed, quality, and cost for voice transcription.
Use Gemini 3 Pro when you need higher quality output for complex or nuanced transcriptions (e.g., technical documentation, multi-speaker content).
Set your default model in Settings → Models, or select per-transcription from the toolbar dropdown.
Model Tiers
| Tier | Model | Use Case |
|---|---|---|
| Standard | Gemini 3 Flash | Daily transcription, notes, emails |
| Premium | Gemini 3 Pro | Complex content, detailed formatting |
Failover Behavior
If the primary model fails, the app automatically retries with the fallback model. By default:
- Primary:
google/gemini-3-flash-preview - Fallback:
google/gemini-3-pro-preview
Configure primary and fallback models in Settings → Models.