Skip to main content

AI Settings of LiveChatAI

LLM Model Selection & Usage

Choose the best large language model (LLM) for your chatbot’s goals. Each model has different strengths, speeds, and credit costs—this guide helps you understand which fits your use case best.

Where to Configure the Model

Where to Configure the Model
  1. Go to your LiveChatAI dashboard
  2. Navigate to Settings → AI Configuration → LLM Model
  3. Select from the available model list
  4. Click Save Changes

Available Models and Use Cases

Model

Recommended Use Case

Message Credit Cost

GPT-4o

In-depth technical support (APIs, code troubleshooting); multi-turn dialogs with nuance

1

GPT-4o-mini

Standard product/service FAQs; balances accuracy with low latency

1

O3-mini

High-volume flows (shipping/account); best for cost-efficient support and prototypes

2

O1

General-purpose support (retail, hospitality); works well for moderately complex queries

20

Claude 3 Sonnet (Beta)

Regulated or sensitive industries (healthcare, finance); policy-compliant interactions

1

Claude 3.5 Sonnet (Beta)

Same as above, with faster performance for medium-length procedures

1

Llama 3.1 70B

Internal knowledge bases; good reasoning at low cost

1

Llama 3.1 405B

R&D use, engineering/scientific support, long-form reasoning

1

Llama 4 Scout 17B

Cost-efficient customer support with solid quality (e-learning, mobile apps)

2

Llama 4 Maverick 17B FP8

Sub-second responses for simple-to-moderate queries in high-traffic environments

1

DeepSeek R1

Real-time dashboard queries, data-driven retrieval from enterprise sources

1

DeepSeek V3

Fast knowledge search across massive document sets (legal, compliance, internal)

1

Gemini 2.5 Pro

Developer support with code and multi-modal help (image + text troubleshooting)

2

Grok-3 (Latest)

Public-facing bots with real-time awareness (news, prices, trending topics)

4

Understanding Message Credit Cost

  • Each “Message Credit” reflects 1 user message + 1 AI reply
  • Higher-performing or proprietary models (like O1) use more credits per exchange
  • Be sure to check your plan’s credit quota when switching models—especially for premium models

Example: A chatbot answering 1 000 user questions with GPT-4o will cost 1 000 credits. The same load on O1 would cost 20 000 credits.

Tips for Choosing the Right Model

  • Need speed + scale? Try O3-mini or Llama 4 Maverick
  • Have a regulated audience? Use Claude 3/3.5 Sonnet
  • Building developer tools? Use Gemini 2.5 Pro or GPT-4o
  • Want real-time awareness? Try Grok-3
  • Working on internal tools? Use Llama 3.1 70B or DeepSeek

For questions or help selecting a model, reach out to [email protected].