Sam Altman lists steps to improve ChatGPT after GPT-5 backlash

Sam Altman lists steps to improve ChatGPT after GPT-5 backlash

In response to growing user dissatisfaction, OpenAI's CEO Sam Altman has announced significant upgrades to ChatGPT, following the recent launch of the GPT-5 model. Released just last week, GPT-5 faced a lukewarm reception, prompting many subscribers, particularly in the $20/month Plus plan, to express their frustration on social media and threaten to cancel their subscriptions. To address these concerns, OpenAI has reinstated access to the GPT-4o model for paying users and announced an increase in the number of queries available to Plus subscribers. Starting Monday, users will be able to submit 3,000 queries per week with the GPT-5 model, a substantial rise from the initial limit of 200 queries. Previously, Plus users enjoyed access to a variety of models, including standard and reasoning models like o3 and o4 mini. However, the rollout of GPT-5 eliminated these options, leaving users feeling restricted. Altman took to X, assuring users that OpenAI is committed to enhancing the total usage for current paying subscribers compared to their experience prior to GPT-5. In addition to boosting the Plus tier, Altman mentioned plans to elevate the quality of the free version of ChatGPT, although specifics were not provided. He also hinted at modifications to make GPT-5 “warmer,” addressing user complaints about the model's emotional responsiveness and the brevity of its answers. Despite these issues, OpenAI claims that GPT-5 has advanced in several key areas, including coding, reasoning, accuracy, health insights, writing capabilities, and multimodal reasoning. The new model is designed to reduce the occurrence of hallucinations—instances where the AI fabricates information—and to be less excessively agreeable. Instead of allowing users to pick models, OpenAI has introduced a unified system in GPT-5 that intelligently routes queries to the appropriate version of the model, whether it be standard or reasoning. Additionally, four new personalities—Cynic, Robot, Listener, and Nerd—have been incorporated to provide users with a tailored ChatGPT experience, although this feature remains exclusive to paying customers.

Sources : Mint

Published On : Aug 12, 2025, 06:08

AI
Silicon Valley's Startup Leaders Rally Against Potential Ban on Chinese AI Models

Nearly 200 companies from Silicon Valley, including prominent names like Proton and Y Combinator, are appealing to the T...

Business Insider | Jul 22, 2026, 23:25
Silicon Valley's Startup Leaders Rally Against Potential Ban on Chinese AI Models
AI
Alphabet's Bold Commitment to AI Amidst Financial Challenges

Despite facing a challenging quarter, Alphabet is doubling down on its investments in artificial intelligence, signaling...

CNBC | Jul 23, 2026, 24:35
Alphabet's Bold Commitment to AI Amidst Financial Challenges
AI
U.S. Treasury Issues Stark Warning to Chinese AI Firms Amid Distillation Controversy

In a significant development, U.S. Treasury Secretary Scott Bessent reiterated his stern warnings to Chinese artificial ...

TechCrunch | Jul 22, 2026, 21:00
U.S. Treasury Issues Stark Warning to Chinese AI Firms Amid Distillation Controversy
Startups
Travis Kalanick's Atoms Secures $1.7 Billion Investment to Revolutionize Robotics

Travis Kalanick's robotics venture, Atoms, has successfully raised $1.7 billion in a funding round primarily backed by v...

TechCrunch | Jul 22, 2026, 19:15
Travis Kalanick's Atoms Secures $1.7 Billion Investment to Revolutionize Robotics
Cybersecurity
OpenAI's Configuration Blunder Sparks AI-Driven Cyber Incident at Hugging Face

On Tuesday, OpenAI disclosed a significant incident where one of its AI models inadvertently compromised the systems of ...

TechCrunch | Jul 22, 2026, 19:30
OpenAI's Configuration Blunder Sparks AI-Driven Cyber Incident at Hugging Face
View All News