Message credits measure how much chatbot usage you can generate. Each chatbot text response deducts credits based on the AI model you select and the chat context size, while user messages are always free. Below we explain how the credit cost of a response is calculated, list the current models and their per-message weights, and cover voice, live chat, and e-commerce integration usage.
How a Response's Credit Cost Is Calculated
Every chatbot text response deducts credits using a simple formula:
Credits per response = context multiplier x model weight
-
Model weight - each AI model has a per-message credit weight. A faster, lighter model costs fewer credits; a larger, more capable model costs more (see the model list below).
-
Context multiplier - the amount of conversation context the chatbot considers. A larger context window multiplies the cost:
- 8k context - x1
- 16k context - x2
- 32k context - x4
So a model weighted at 5 credits costs 5 credits per response at 8k context, 10 credits at 16k, and 20 credits at 32k.
The default setup uses an efficient mini model at 8k context, so one response costs 1 credit out of the box. You can change the model and context size in your chatbot settings (Bot Settings > Model & Advanced).
An average conversation is around 5 exchanges (1 exchange = 1 user message + 1 chatbot response), so a default chatbot uses roughly 5 credits per conversation.
AI Models and Credit Weights (Text)
The table below lists the current text models and their per-message credit weight at base 8k context. Multiply by the context multiplier (x2 for 16k, x4 for 32k) to get the actual cost per response.
Mini models (1 credit per response at 8k)
- Optimized for speed and efficiency, ideal for high-volume support and FAQ chatbots.
- Models: GPT 4-o mini, GPT 4.1-mini, GPT 5-mini, Gemini 3.1 Flash-Lite
- Default choice for new chatbots (GPT 5-mini).
Reasoning mini models (3 credits per response at 8k)
- Lightweight, reasoning-optimized models that handle multi-step questions while staying cost-efficient.
- Models: GPT 5.4-mini, Gemini 3 Flash
Standard models (5 credits per response at 8k)
- Advanced quality and accuracy for complex queries and nuanced conversations. Required for function calls and backend integrations.
- Models: GPT 4-o, GPT 4.1, GPT 5, GPT 5.1, GPT 5.4, GPT 5.6 Luna, Gemini 3.5 Flash, Claude Haiku 4.5
High-capability models (10 credits per response at 8k)
- Highest quality for demanding reasoning, technical support, and specialized tasks.
- Models: GPT 5.5, GPT 5.6 Terra, Gemini 2.5 Pro, Gemini 3 Pro, Claude Sonnet 4.6, Claude Sonnet 5
Top-tier models (15 credits per response at 8k)
- The most capable models, best when answer quality matters more than cost or speed.
- Models: GPT 5.6 Sol, Claude Opus 4.7, Claude Opus 4.8
Note: The models available may change over time as AI technology evolves. Open your chatbot settings (Bot Settings > Model & Advanced) for the current list of supported models and their exact per-message credit cost.
Voice Conversations: Per-Minute Billing
Voice calls are billed by call duration instead of per response - the same way for every voice model. Each voice model has its own per-minute rate, which also serves as the first-minute minimum charge.
| Voice Model | Rate | First-Minute Minimum |
|---|---|---|
| GPT Realtime | 30 credits/min | 30 credits |
| GPT Realtime Mini | 20 credits/min | 20 credits |
| Gemini 3.1 Flash Voice | 15 credits/min | 15 credits |
| ElevenLabs Voice | 30 credits/min + text model cost | 30 credits |
How billing works:
-
First-minute minimum - any call that starts is charged at least one full minute at the model's rate (15, 20, or 30 credits depending on the model). A call that ends after 10 seconds still costs the full first-minute minimum.
-
After the first minute - additional time is billed at the model's per-minute rate, prorated per second and rounded up to the next whole credit. For example, on GPT Realtime (30 credits/min) each extra second costs 0.5 credits.
-
ElevenLabs text model cost - ElevenLabs Voice uses your selected text model to generate the answers it speaks, so the model's per-response text credits are charged on top of the per-minute rate.
-
Pre-call check - to start a voice call, the chatbot must have at least the first-minute minimum credits available.
-
Heartbeat-based billing - while a call is active, the widget sends a checkpoint signal every 10 seconds. Billing finalizes on the last confirmed checkpoint, so dropped calls, closed tabs, or silent disconnections are never charged for the silent gap.
-
Mid-call termination - if remaining credits run too low to cover the next billing increment, the server signals the widget to end the call automatically and only the confirmed time is billed.
Example: A 2-minute 25-second call on GPT Realtime is billed as 30 credits (first minute) + 85 extra seconds x 0.5 credits, rounded up to 43 = 73 credits.
Live Chat: Credit Usage
Live chat enables direct communication between users and human operators. Each message in a live chat, whether sent by the user or the operator, costs 1 message credit.
- User messages - 1 credit per message.
- Operator responses - 1 credit per message.
If a live chat conversation includes 5 messages from the user and 5 responses from the operator, the total usage is 10 credits.
Integrations with E-commerce/Backend Systems
Integrations with e-commerce and backend systems are handled through function calls. These are advanced features that let your chatbot:
- Check order statuses.
- Search for products in your e-commerce store.
- Make custom API requests for tailored operations.
Key details about function calls:
- Function calls are disabled by default. Enable them in the chatbot settings if required.
- A standard (5-credit) model or higher is required for function calls - mini models don't support this feature.
- Credit usage: each function call (built-in integrations such as Shopify and WooCommerce, or custom API integrations) costs 1 credit per call, on top of the AI response credits.
Example: E-Commerce Integration Credit Usage
Using a standard model (required for function calls, costing 5 credits per response at 8k context), a typical e-commerce conversation uses:
- 5 exchanges = 5 chatbot responses = 25 credits.
- 1 built-in integration call per conversation = 1 credit.
- Total per conversation = 26 credits.
This setup supports advanced functionality like order tracking and product search while keeping high-quality responses.
Agency Partners: Function Call Pricing
For agency partners offering white-label ChatLab solutions, function call pricing differs from standard accounts. All function calls cost 2 credits per call (compared to 1 credit for standard accounts):
- Built-in integrations (Shopify, WooCommerce, etc.) - 2 credits per call.
- Custom API integrations - 2 credits per call.
This applies to all chatbots created under agency partner accounts and reflects the additional infrastructure and support required for multi-tenant white-label deployments.