California: OpenAI has opened the curtain on GPT-4.5, its new large language model (LLM) that promises more natural, reflective conversations with far fewer mistakes. Available now to ChatGPT Pro subscribers at $200/month and developers on paid API tiers, this new model marks a major upgrade—albeit at a premium.
CEO Sam Altman praised GPT-4.5 as “the first model that feels like talking to a thoughtful person,” citing its greater capacity to infer user intent and provide more emotionally intelligent answers. In contrast with the new chain-of-thought models from rival companies, GPT-4.5 maintains a more classic LLM method, but also reaps the benefits of leading-edge unsupervised learning methodologies that fuel better performance.
A major hurdle for OpenAI has been GPU shortages. “We’re out of GPUs,” Altman admitted on X, explaining that high demand forced a limited rollout. The company plans to add tens of thousands of GPUs next week, with hundreds of thousands more on the horizon, before gradually expanding access to lower-tier subscriptions like Plus and Team, priced at $20 and $30/month respectively.
GPT-4.5 isn’t just about better conversation— it’s designed for greater reliability. By bringing down hallucination rates from 61.8% in GPT-4o to 37.1%, the model promises to redefine accuracy in human-like AI interactions. New features such as embedded search, ChatGPT Canvas mode, and the facility to upload images and files are also available for users. Features such as voice, video, and screen-sharing, however, are still on the development pipeline.
This release underlines OpenAI’s commitment to pushing the boundaries of AI, blending advanced training techniques with real-world usability—even if it comes at a higher cost.
Stay tuned as the rollout expands and reshapes the landscape of intelligent conversation.
Disclaimer: The news articles published on Growing Pro Technologies are based on reports from reputable third-party sources and are not original reporting by Growing Pro Technologies. While we strive to ensure accuracy and integrity, we cannot guarantee the completeness or timeliness of the information provided
Interesting Reads:
OpenAI Scraps o3, Fast-Tracks GPT-5 for a Unified AI Push
DeepSeek’s AI Breakthrough Sends Shockwaves Through Tech Markets
