OpenAI to Release GPT-5 This Summer

OpenAI is set to release the next version of its ChatGPT model, GPT-5, by mid-year, likely during summer. Enterprise customers have received demos of the latest model and its enhancements. A CEO who saw GPT-5 described it as "materially better" and mentioned new capabilities, including AI agents that perform tasks autonomously.

The release date is not fixed as OpenAI is still training GPT-5. After training, it will undergo internal safety testing and "red teaming" to identify issues. This process could delay the release.

Sales to enterprise customers are OpenAI's main revenue stream. GPT-4, released a year ago, was faster and more accurate than GPT-3. Later, GPT-4 Turbo was introduced to address issues like "laziness."

Large language models are trained on vast data sets from the web, which can affect the quality of responses. Complaints about GPT-4's degradation have surfaced on Reddit. OpenAI employees hope GPT-5 will be more reliable and impressive.

Copyright issues remain a challenge as much of the crucial training data is owned by copyright holders. OpenAI and other tech companies are advocating against updated federal rules for LLMs' access to such material.

For tips or insights, contact Kali Hays at khays@businessinsider.com or on Signal at 949-280-0267. Reach Darius Rafieyan on Signal at +1-714-651-1367 or email drafieyan@businessinsider.com.

The original article: https://www.businessinsider.com/openai-launch-better-gpt-5-chatbot-2024-3