Model Fine Tuning lets you create a custom-trained version of ChatGPT that follows your rules and responds the way you want. You write example conversations showing the model what a user would say and what the correct response should be, and the training process teaches the model to follow those patterns. Once trained, your custom model is available in the AI Chatbot and AI Bots apps as a model option, so any chatbot or AI bot you build can use it.

Fine-tuning works best for repeatable, non-complex tasks where you want consistent behavior, customer support responses, specific formatting styles, domain-specific terminology, or any situation where the model should follow a predictable pattern. It is not a replacement for powerful reasoning models. Think of it as training an employee: you show them how to handle common situations, and they learn to respond the same way every time.

The strongest model OpenAI allows for fine-tuning is GPT-4o-mini, which handles conversational tasks well at low cost per message. You train it by writing conversation blocks in the admin panel. Each block contains a prompt (what the user says) and the correct response (what the model should say back), with up to 3 exchanges per block for multi-turn conversations. Set your global rules first. These are injected into every training message so the model learns consistent behavior from the start. When your custom model is used in a chatbot or AI bot, those same rules are loaded into the system prompt so the model operates under the same guidelines it was trained on.

All training is managed through the admin panel. Write your training messages, set your global rules, and run the training job. The system uploads your examples to OpenAI, creates the fine-tuning job, and tracks its progress. Once complete, the custom model name is stored and automatically appears in the model selection for AI Chatbot and AI Bots. You can continue adding more training messages and retrain to improve the model over time.

• Custom train a GPT-4o-mini model on your own conversation examples to teach it your style, rules, and expected responses.
• Global rules that apply to every training message so the model learns consistent behavior across all examples.
• Multi-turn training conversations with prompts and pre-selected correct responses, up to 3 exchanges per block.
• Trained models automatically appear in the AI Chatbot and AI Bots model selectors, no manual configuration needed.
• When your custom model is used in a chatbot or AI bot, its trained rules are loaded into the system prompt automatically.
• Tag and organize training messages by category to keep your training data structured.
• Re-train individual messages by resetting their status, or clear messages after training to save storage.
• Training status tracking with the ability to cancel jobs in progress.
• Uses your own OpenAI API key for all training, you control the costs and the resulting model.

Software Fee: 1 dollar per training session, plus 1 credit per message trained. Database storage credits apply for messages kept in their database. Messages can be cleared after training, or kept for reuse.