Most impressive thing:
The token limits goes from 4K up to 32K tokens (or roughly 50 pages of texts).
Now we can use GPT4 to write long novel.
Other important takeaways of GPT4:
- GPT-4 is more reliable, creative, and capable of handling more nuanced instructions than GPT-3.5, especially for complex tasks.
- GPT-4 can accept both text and image inputs and generate text outputs in various domains.
- Steerability allows API users to customize their AI’s style and task by describing the directions in the “system” message.
- GPT-4 still has limitations, including hallucinating facts and making reasoning errors, and language model outputs should be used with caution, especially in high-stakes contexts.
- GPT-4 significantly reduces hallucinations compared to previous models and scores 40% higher than GPT-3.5 on internal adversarial factuality evaluations.