GPT-4.5 Arrives for ChatGPT Plus Subscribers—Here's What to Expect
OpenAI has introduced GPT-4.5, its latest and most extensive AI model to date, signaling a significant advancement in the competitive AI landscape. Subscribers on the ChatGPT Plus plan will gradually receive access to this new model over the next one to three days. CEO Sam Altman disclosed that initial access will be limited to manage user expectations due to capacity constraints.
Features of GPT-4.5
GPT-4.5 brings the promise of enhanced conversational skills and a deeper understanding of the world, aiming to reduce inaccuracies, also known as “hallucinations,” that were prevalent in earlier models. Distinguished by its substantial increase in data and computational resources during training, GPT-4.5 stands as OpenAI’s most substantial model. The company highlights its proficiency in practical problem-solving, casual chat interactions, and creative assignments.
Despite its scale and intensive training, GPT-4.5 may not universally outperform its predecessors. OpenAI has acknowledged that when it comes to intricate reasoning tasks, GPT-4.5 falls short compared to newer models, trailing behind competitors like Anthropic’s Claude and DeepSeek’s AI models. Sam Altman emphasized this aspect, stating, “GPT-4.5 isn’t a reasoning model and won’t excel in benchmarks.”
Price Considerations
The pricing of GPT-4.5 raises eyebrows, priced at $150 per million generated tokens, making it 15 times more costly than OpenAI’s popular GPT-4o model. This considerable price increase invites speculation about its widespread adoption, particularly considering its limitations in logical reasoning and multi-step problem-solving capabilities.
Advantages of GPT-4.5
Notwithstanding these concerns, GPT-4.5 boasts compelling features. Internal tests by OpenAI demonstrate its effectiveness in persuading other AI models, showcasing its persuasive conversational abilities. Moreover, it offers comprehensive support for ChatGPT tools and APIs, granting developers extensive flexibility.
Conclusion
As GPT-4.5 becomes accessible to users in the coming days, OpenAI’s release reflects its ongoing exploration of the trade-offs between power, performance, and practical utility. Users anticipating the latest model will soon have the chance to test it, while OpenAI continues its evaluation of how far and how quickly they can advance their technology.
