GPT-4o (openai.com)

Article summary

The discussion revolves around OpenAI's new update, GPT-4o, which features real-time audio and multimodal capabilities. The update is seen as a significant improvement, but some users are disappointed that it does not bring a significant increase in intelligence. The new model is being rolled out to free users, raising questions about the value of the paid ChatGPT Plus subscription. The update also includes a new tokenizer with a larger vocabulary size, which is expected to improve efficiency and reduce costs.

Main themes

• GPT-4o update
• Multimodal capabilities
• ChatGPT Plus value
• Tokenizer improvements
• Real-time audio
• Intelligence updates

What commenters say

• The new GPT-4o update is a significant improvement, but it does not bring a significant increase in intelligence, which is a disappointment for some users.
• The multimodal capabilities of GPT-4o, including real-time audio, are a game-changer and make the model more useful for various applications.
• The value of the paid ChatGPT Plus subscription is questionable now that the new model is being rolled out to free users, but some argue that future updates will justify the cost.
• The new tokenizer with a larger vocabulary size is expected to improve efficiency and reduce costs, especially for non-English languages, which previously required more tokens to represent the same text.
• Some users are skeptical about the new update and worry that it may worsen the completion quality to make it cheaper for OpenAI to run.
• The desktop app with voice and vision capabilities is a welcome addition, but some users are disappointed that it is only available for Mac users.
• The update's focus on multimodal capabilities is seen as a gimmick by some, while others believe it is a significant improvement that will increase usability and make the model more useful.
• The new model's capabilities will be rolled out over the next few weeks, and some users are excited to see how it will develop and improve.