11/07/2023
OpenAI's latest marvel, GPT-4 Turbo, is coming. Here are the 6 fresh highlights from OpenAI:
1. Mega Context Length:
GPT-4 Turbo can handle a massive 128k tokens - imagine the content of a 300-page book fitting into a single prompt!
2. Enhanced Control:
Enjoy a new JSON mode, superior function calling capabilities, multi-function calls, and better adherence to instructions. Plus, there's a new beta feature for reproducible outputs, allowing consistent results with a seed parameter.
3. Updated Knowledge:
Stay current with knowledge up to April 2023, and integrate external documents or databases seamlessly into your projects.
4. Exciting New Features:
Dall-E 3, GPT-4 Turbo with vision capabilities, and a brand-new text-to-speech model are all hitting the API today. Plus, GPT-4 Turbo now supports image inputs - a feature bound to inspire incredible creations. And, the next version of Whisper (V3) is on its way!
5. Advanced Customization:
Fine-tuning is now accessible for the 16k version of GPT-3.5. Applications are open for GPT-4 fine-tuning access, and there's a bespoke Custom Models program for enterprise-level model personalization.
6. Increased Rate Limits:
All established GPT-4 customers will enjoy a doubled token rate per minute.
via openai
