OpenAI today unveiled GPT-4 Turbo, a new version of its flagship product that promises to be a real breakthrough in the field of artificial intelligence. This version includes a number of significant improvements:
- Extended context: GPT-4 Turbo can now process up to 128,000 tokens, which is equivalent to about 300 pages of a book. This allows the AI to analyze and generate much longer texts.
- More control: In the new API version, GPT-4 can respond in valid JSON format and call multiple functions, making it more flexible for developers.
- Improved knowledge: The built-in information retrieval mechanism with data relevance as of April 2023 significantly enhances the quality and relevance of responses.
- Multi-modality: GPT-4 Turbo is integrated with DALLE, text-to-speech features, and visual perception, expanding the capabilities of AI.
- Image input via API 😱: The API can now process incoming images, opening up new possibilities for visual analysis and content generation.
- Sound generation: The text-to-speech feature allows for the creation of realistic audio content.
- Experimental access to fine-tuning: OpenAI is opening access to fine-tuning GPT-4, providing developers with even more tools for customizing the AI.
- Custom models: OpenAI offers collaboration with companies to create specialized models for specific business tasks.
Moreover, GPT-4 Turbo will cost three times less than the previous version, GPT-4.
🔗 Link: https://openai.com/blog/new-models-and-developer-products-announced-at-devday
