GPT-4 API General Availability (openai.com)

Article summary

The article announces the general availability of the GPT-4 API. The discussion focuses on the comparison between GPT-3.5 and GPT-4, with some users finding GPT-3.5 sufficient for their needs, while others see significant quality differences between the two models. The price increase of GPT-4 is also a point of consideration for many users. The deprecation plan for older models of the Completions API is also mentioned.

Main themes

• GPT-3.5 vs GPT-4 comparison
• Price and cost considerations
• Domain-specific performance
• Deprecation of older models
• Programmatic access and fine-tuning
• Use case dependence

What commenters say

• GPT-3.5 is sufficient for many use cases, especially with system prompt engineering, and the price increase of GPT-4 may not be justified.
• GPT-4 is significantly better than GPT-3.5 in certain domains, such as legal writing and code completion.
• The quality difference between GPT-3.5 and GPT-4 can be highly variable depending on the specific use case and domain.
• Some users find GPT-3.5 to be unreliable and prone to errors, while others see it as a viable alternative to GPT-4.
• The deprecation of older models of the Completions API is seen as a step back by some users, who prefer the flexibility of the completion API over the chat API.
• Programmatic access to GPT-3.5 can be beneficial in certain scenarios, allowing for faster and cheaper responses.
• Fine-tuning a model on a specific domain can lead to significant improvements in performance.
• The choice between GPT-3.5 and GPT-4 ultimately depends on the specific needs and use case of the user.