Sam Altman details steps to improve ChatGPT as users threaten to cancel OpenAI subscriptions – Crypto News – Crypto News
Connect with us
Sam Altman details steps to improve ChatGPT as users threaten to cancel OpenAI subscriptions Sam Altman details steps to improve ChatGPT as users threaten to cancel OpenAI subscriptions

Technology

Sam Altman details steps to improve ChatGPT as users threaten to cancel OpenAI subscriptions – Crypto News

Published

12 months ago

on

OpenAI CEO Sam Altman on Tuesday confirmed a number of major changes coming to ChatGPT owing to increased demand after the GPT-5 launch. The ChatGPT maker had released its latest language model last week, but the reception to that model was far from ideal, with many OpenAI paying customers on social media threatening to cancel their subscriptions owing to the removal of all other previous AI models.

Since then, OpenAI has taken various steps to woo these users, especially the ones in its $20/month Plus tier. OpenAI had earlier brought back GPT-4o for paid users and, on Monday, said that it was planning to provide 3,000 queries per week of the GPT-5 thinking model to Plus users, up from 200/week at the launch of GPT-5.

The major pain point for ChatGPT Plus users has been that access to multiple models prior to the GPT-5 launch gave them a lot of queries to work with, including the standard GPT models and reasoning models like o3 and o4 mini. However, with the GPT-5 launch, all of these models were stripped away, and users were left with only GPT-5 to work with.

Sam Altman announces changes coming to ChatGPT

In a post on X on Tuesday, Altman assured that OpenAI will “make sure that current paying ChatGPT users get more total usage than they did before GPT-5.”

Altman further stated that OpenAI will “increase the quality of the free tier of ChatGPT” but did not give any specifics on what that will entail.

The OpenAI CEO had earlier said that the company would make changes to GPT-5 in order to make the model “warmer.” Notably, a major complaint from users with GPT-5 has been that the model lacks emotional depth in its answers compared to the previous AI models, while also giving much shorter answers.

What’s new with GPT-5?

OpenAI says that GPT-5 makes big strides in many areas over its predecessors, including coding, reasoning, accuracy, health, writing, and multimodal reasoning. The model is said to encounter fewer hallucinations (making stuff up) and be less sycophantic (being too agreeable with users).

With GPT-5, OpenAI also removed the model picker and instead released a new unified system with a real-time router choosing which queries to send to the GPT-5 standard model and which ones to send to the reasoning model.

OpenAI has also brought four new personalities with GPT-5 — Cynic, Robot, Listener, and Nerd — to allow users to customise their experience with ChatGPT. This feature, however, is restricted to OpenAI’s paying customers.

Related Topics:
Continue Reading

Trending