# Provider Costs Breakdown

> Detailed costs by provider

Source: https://edesy.in/docs/voice-agent/billing/provider-costs

Detailed pricing for each provider (approximate, per minute).

## STT Providers

| Provider | Cost/min | Best For |
|----------|----------|----------|
| Deepgram | Lowest | English, speed |
| OpenAI Whisper | Low | Quality |
| ElevenLabs Scribe | Low | Multilingual |
| AssemblyAI | Medium | English |
| Azure | Medium | Enterprise |
| Google Chirp | Higher | Accuracy |

## TTS Providers

| Provider | Cost/min | Best For |
|----------|----------|----------|
| Deepgram | Lowest | Speed |
| Sarvam | Low | Indian languages |
| Cartesia | Low | Low latency |
| Azure | Medium | Regional languages |
| OpenAI | Medium | Multilingual |
| Google | Medium | General |
| ElevenLabs | Premium | Highest quality |

## LLM Providers

| Provider | Cost/min | Best For |
|----------|----------|----------|
| Gemini 2.5 Flash-Lite | Lowest | Real-time voice |
| Gemini 2.0 Flash | Low | Indian languages |
| GPT-4o-mini | Low | Budget quality |
| Claude 3 Haiku | Low | Fast |
| GPT-4o | Medium | Best reasoning |
| Claude 3.5 Sonnet | Medium | Complex tasks |

## Native Audio Models

| Model | Cost/min | Notes |
|-------|----------|-------|
| Gemini Live 2.0 | Low | Standard |
| Gemini Live 2.5 HD | Low | Premium features |
| OpenAI Realtime Mini | Lowest | Budget |
| OpenAI Realtime | Medium | Premium |

## Cost Optimization Tips

### Budget-Conscious Setup

```
STT: Deepgram
LLM: Gemini 2.5 Flash-Lite
TTS: Sarvam or similar
────────────────────────────────────
Total: Very low cost per minute
```

### Quality-First Setup

```
STT: ElevenLabs Scribe
LLM: GPT-4o
TTS: ElevenLabs
────────────────────────────────────
Total: Higher cost, premium quality
```

### Native Audio (Lowest Latency)

```
Gemini Live 2.5 HD
(Includes voice input and output)
Lowest latency, competitive cost
```