Do you want smarter insights in your inbox? Sign up for our weekly newsletters to get what is critical for AI leaders, data and security. Subscribe now
Openai once again creates a GPT-4-duty language model (LLM), which powered CHATGPT before starting the GPT-5 last week-a home option for all paying users, that is, those who subscribe to chatgPT plus (USD 20 per month), PRO (USD 200 per month), team (30 USD per month), enterprises or edu. Do not require users to switch on the “Show older models” to access it.
However, paying CHATGPT subscribers will also receive fresh “show additional models” by default, which restores access to GPT-4.1, O3 and O4-Mini, the last two LLM-oriented reasoning.
OPENAI CEO and co -founder Altman himself announced a change to X just a few minutes agopromising that if the company ever removes GPT-4O in the future, It will give “a lot of attention”.
Models can be found in the “Picker” menu at the top of the ChatGPT session screen on the Internet and on mobile and other devices.
AI scaling hits its limits
Power capitals, the growing costs of the token and inference delay are transforming AI Enterprise. Join our exclusive salon to discover how the best teams are:
- Changing energy into a strategic advantage
- Architect of effective inference regarding real capacity profits
- Unlocking competitive roi using balanced AI systems
Secure your place to remain ahead: https://bit.ly/4mwgni
The reversal occurs after a turbulent first week for the GPT-5, which developed on August 7 in four variants-regular, mini, nano and pro-Z optional “thinking” modes in several of them longer, more demanding reasoning.
As previously reported by Venturebeat, the GPT-5 debut met with mixed reviews and infrastructure hiccups, including the broken “Autoswitcher”, which managed the error incorrectly, inconsistent performance compared to GPT-4O and the user’s frustration regarding the sudden removal of older models.
The latest Altman update adds fresh control elements to the ChatGPT interface: Users can now choose between “Auto”, “Fast” and “thinking” modes for GPT-5.
“Thinking” mode-from 196,000 in the context-like window-the now contains 3000 messages per week for paying subscribers, after which they can continue using the lighter “Mini GPT-5 thinking” mode. Altman noticed that borders can change depending on the trends of exploit.
However, GPT-4.5 remains only for Pro users due to the high GPU costs.
Altman also suggested another change on the horizon: a personality correction for the GPT-5, which is to feel “warmer” than the current by default, but less polarizing than tons of GPT-4O.
The company explores the user adaptation as a long-term solution-which can solve forceful emotional attachments that some users have created with specific models.
For now, changes should facilitate composed users who felt frustrated with a sudden transition to GPT-5 and the depreciation of older LLM OPENAI, although it can also still drive intensive emotional fixations that some users have developed with these models.
