> ## Documentation Index
> Fetch the complete documentation index at: https://docs.talkifai.dev/llms.txt
> Use this file to discover all available pages before exploring further.
# Billing & Credits
> Understand how TalkifAI billing works, manage your credits, and optimize costs across voice and text agent usage.
## Overview
TalkifAI uses a **credit-based billing system**. You purchase credits, and credits are consumed as your agents handle conversations. This gives you predictable costs and full control over spending.
**Key points:**
* Credits are pre-purchased (no surprise monthly bills)
* Credits never expire
* Usage tracked per minute for voice, per token for text
* All billing is per-organization
***
## How Credits Work
```
Buy Credits → Deploy Agents → Agents Handle Calls → Credits Consumed
```
Credits are consumed based on:
| Usage Type | Credit Rate |
| -------------------------- | --------------------------------- |
| **Voice Call** | Per minute of active conversation |
| **Text Chat** | Per 1,000 tokens (input + output) |
| **Post-Call Analysis** | Per analysis run |
| **Knowledge Base Queries** | Per similarity search |
Exact credit rates are shown in your **Dashboard → Billing** page. Rates vary based on your plan and selected AI providers.
***
## Billing Modes
The cost of running an agent depends on your **Key Mode** (set in Settings → API Keys):
### TalkifAI Keys Mode (Default)
TalkifAI covers all provider costs (OpenAI, Deepgram, Cartesia, Google). You pay a **single per-minute rate** that includes everything.
**Best for:** Simplicity. One rate, one invoice.
```
You pay TalkifAI Credits
└── TalkifAI pays: OpenAI + Deepgram + Cartesia
```
### My Keys (BYOK) Mode
You provide your own provider API keys. You pay TalkifAI a **reduced platform fee** only. Provider costs are billed directly to your provider accounts.
**Best for:** High-volume deployments where you want direct provider relationships and potentially lower costs.
```
You pay TalkifAI Credits (platform fee only)
+
You pay OpenAI directly (LLM costs)
You pay Deepgram directly (STT costs)
You pay Cartesia directly (TTS costs)
```
***
## Purchasing Credits
### Via Dashboard
1. Go to **Settings → Billing**
2. Click **Add Credits**
3. Select a credit package or enter a custom amount
4. Complete payment via credit card
5. Credits are added instantly to your account
### Credit Packages
| Package | Credits | Price |
| ---------- | ------- | --------------- |
| Starter | 1,000 | \$10 |
| Growth | 5,000 | \$45 (10% off) |
| Scale | 20,000 | \$160 (20% off) |
| Enterprise | Custom | Contact us |
***
## Auto Top-Up
Prevent service interruptions by enabling automatic credit top-up:
1. Go to **Settings → Billing**
2. Enable **Auto Top-Up**
3. Set a **Low Balance Threshold** (e.g., 500 credits)
4. Set a **Top-Up Amount** (e.g., 2,000 credits)
5. Save your payment method
When your balance drops below the threshold, TalkifAI automatically purchases the configured amount and charges your saved payment method.
Auto top-up uses your **saved payment method**. Update it in Settings → Billing → Payment Methods.
***
## Monitoring Usage
### Credit Balance
Your current balance is shown in the **top navigation bar** of Studio. Click it to go to the Billing page.
### Usage Breakdown
Go to **Settings → Billing → Usage** to see:
* Credits consumed per day / week / month
* Breakdown by agent
* Breakdown by call type (voice, text, analysis)
* Export usage as CSV
### Usage Alerts
Set up alerts to notify you before credits run out:
1. Go to **Settings → Billing → Alerts**
2. Enable **Low Balance Alert**
3. Set threshold (e.g., "Alert me when below 1,000 credits")
4. Enter notification email
5. Save
***
## Developer Quotas
For **developer accounts** within your organization, you can set monthly minute limits to control individual usage:
* Default quota: **3,000 minutes/month** per developer
* Check current usage: **Analytics → Usage by Member**
* Adjust quota: **Settings → Team → \[Member] → Edit Quota**
When a developer reaches their quota, their calls are blocked until the next month or until their quota is increased by an admin.
***
## Cost Optimization
### Choose the Right Architecture
| Architecture | Relative Cost | Best For |
| ---------------------------- | ------------- | ----------------------------- |
| **Realtime (OpenAI/Gemini)** | Higher | Low-latency priority |
| **Pipeline** | Moderate | Balance of cost + flexibility |
| **Text Chat** | Lowest | Non-voice use cases |
### Choose Cost-Effective Providers
When using Pipeline architecture, mixing providers affects cost:
```
Most Expensive: OpenAI GPT-4o + ElevenLabs + Deepgram
Most Balanced: GPT-4o-mini + Cartesia + Deepgram
Most Affordable: Gemini Flash + Cartesia + Deepgram
```
### Optimize Post-Call Analysis
Post-call analysis adds to cost. Optimize by:
* Use **GPT-4o-mini** or **Gemini Flash** (cheaper than GPT-4o)
* Limit analysis to **5-7 fields** (fewer = cheaper)
* Analyze **100% for compliance**, sample 10-20% for general QA
### Reduce Call Duration
Shorter calls = fewer credits. Improve agent efficiency:
* Write concise system prompts (agents that talk less)
* Enable **inactivity timeout** to end abandoned calls
* Use effective triage to route calls faster
***
## Invoices & Billing History
### Download Invoices
1. Go to **Settings → Billing → History**
2. Click **Download Invoice** next to any purchase
3. Invoices are in PDF format for accounting
### Billing Email
By default, receipts go to the organization owner's email. Update:
1. Go to **Settings → Billing**
2. Under **Billing Email**, enter the desired address
3. Save
***
## Enterprise & Custom Plans
For high-volume deployments or custom requirements:
* **Volume discounts** on large credit purchases
* **Monthly invoicing** instead of pre-purchase
* **Dedicated support** SLA
* **Custom rate limits** for batch calling
* **SLA guarantees** for uptime
Contact [support@talkifai.dev](mailto:support@talkifai.dev) or your account manager for enterprise pricing.
***
## Troubleshooting
**Problem:** Agents stop working due to low balance.
**Solutions:**
1. Purchase additional credits in Settings → Billing
2. Enable Auto Top-Up to prevent future outages
3. Set a Low Balance Alert to be notified early
**Check:**
1. Go to Settings → Billing → Usage to see which agents/call types are consuming most
2. Look for unusually long calls (inactivity timeout may be off)
3. Check if Post-Call Analysis is running on all calls unnecessarily
4. Verify no test agent is running in an infinite loop
**Quick wins:** Enable inactivity timeout, switch to cheaper LLM model for low-stakes agents.
**Common causes:**
* Card declined by bank (3D Secure / international transaction)
* Card expired
* Insufficient funds
**Fix:**
1. Update payment method in Settings → Billing → Payment Methods
2. Try a different card
3. Contact your bank if card is being declined
**Contact support:** Email [support@talkifai.dev](mailto:support@talkifai.dev) with your invoice number. Include the dates in question and your usage CSV export from the billing dashboard.
***
## Next Steps
Purchase credits to start deploying agents.
Never run out of credits with automatic balance management.
Review Analytics to find cost reduction opportunities.
Switch to BYOK mode to reduce per-minute costs at scale.