A new program allows users to create custom models fine-tuned for specific domains, enhancing the capabilities of chat interactions. The token limit for the API has doubled, enabling developers to process more data efficiently. Additionally, a new voice feature, Whisper, will be open source, allowing users to interact with AI without ongoing costs. Notably, the pricing for GPT-4 Turbo has been reduced by three times, making it more accessible for integration into various applications.