Providers and multipliers
Complete table of transcription, translation and AI providers. The multiplier indicates how many minutes of your quota each provider consumes — pricier providers consume it faster.
Transcription providers (ASR)
The multiplier affects plan quota consumption. Base 1.0x = Whisper Turbo. A 2.0x provider consumes twice the minutes.
| Provider | Mode | Multiplier | Note |
|---|---|---|---|
| Alibaba Paraformer-v2 | Recorded | 0.5x | Ultra cheap for Chinese |
| SenseVoice (self-hosted) | Recorded | 0.6x | Variable GPU cost |
| Soniox v4 | Recorded | 0.8x | Multilingual with PT-BR support |
| Alibaba Qwen3-ASR | Recorded | 1.0x | Asia/Global |
| Whisper V3 Turbo | Recorded | 1.0x | Reference baseline (1.0x) |
| Voxtral (Mistral) | Recorded | 1.3x | 13 languages, native diarization |
| Soniox v4 | Live | 0.9x | Two-way translation |
| AssemblyAI U-2 | Recorded | 1.1x | 99 languages |
| Gladia | Recorded | 3.8x | Code-switching, 100+ languages |
| ElevenLabs Scribe v2 | Recorded | 1.5x | Editorial-grade text |
| AssemblyAI U-3 Pro | Recorded | 1.5x | Premium multilingual |
| Gladia | Live | 4.6x | WebSocket, code-switching |
| ElevenLabs Scribe | Live | 2.5x | Browser-direct |
| AssemblyAI U-3 Pro | Live | 2.8x | Premium streaming |
| Deepgram Nova-3 | Recorded | 2.9x | Specialized models |
| Deepgram Nova-3 | Live | 2.9x | Nova-3 streaming |
| Recall.ai (Meeting Bot) | Live | 2.9x | Meeting bot |
| iFlytek | Recorded | 4.3x | Specialized in Chinese |
| iFlytek | Live | 4.8x | zh→en translation |
The multiplier defines how much of your quota each provider consumes. The final price is your plan (included in quota) or overage.
Translation providers
Translation as an add-on. The points consumed vary by the selected engine.
| Provider | Points/minute | Note |
|---|---|---|
| Gemini Flash Lite | 200 pts/min | Low-cost baseline option |
| GPT-5.6 Luna | 350 pts/min | Intermediate |
| Azure Translator | 250 pts/min | Enterprise |
| Google TLLM | 300 pts/min | Gemini-based, glossary |
| Google NMT | 500 pts/min | 135 languages |
| DeepL | 800 pts/min | Option for European languages |
AI features
Points consumed per use. Points are deducted from the plan monthly quota.
| Feature | Points | Unit |
|---|---|---|
| Chat about transcript | 500 pts | per message |
| Knowledge Base (RAG) | 300 pts | per query |
| Follow-up email | 800 pts | per job |
| Repurpose (blog, LinkedIn) | 1,000 pts | per job |
| Flashcards | 1,500 pts | per job |
| Quiz / exercises | 1,500 pts | per job |
| Executive summary | 2,000 pts | per job |
| Mind map | 2,000 pts | per job |
| Study guide | 2,000 pts | per job |
Multiplier by LLM model
More powerful models consume more points. The multiplier applies over the base cost (Gemini Flash Lite = 1.0x).
| Model | Multiplier | Note |
|---|---|---|
| Gemini 3.1 Flash Lite | 1.0x | Product reference baseline |
| GPT-5.6 Luna | 2.5x | Intermediate |
| Gemini 3.5 Flash | 2.8x | Fast and accurate |
| Claude 4.5 Haiku | 5.6x | Fast, high quality |
How consumption works
- • Transcription: each provider has a multiplier. 1 hour with Soniox (0.8x) consumes 48 min of quota. 1 hour with Deepgram (2.9x) consumes 174 min.
- • Translation: consumes points according to the chosen engine; premium engines consume more points.
- • AI (summary, chat, etc.): consumes points from the monthly quota. More powerful models (Claude Haiku, Gemini 3.5 Flash) consume more points.
- • Overage: when the quota runs out, it charges per minute according to the plan (Starter $0.07, Pro $0.05, Business $0.035/min).