OpenAI has announced the launch of GPT-4.5, code-named Orion, which is its largest AI model to date, utilizing unprecedented computing power and data. Despite its size, OpenAI’s white paper clarifies that GPT-4.5 is not considered a frontier AI model. Subscribers to ChatGPT Pro and developers on paid API tiers will gain access to GPT-4.5, while other users will receive it shortly after. The model exhibits improved performance in various benchmarks but shows signs of diminishing returns compared to previous generations, particularly against newer reasoning models from competitors.
While GPT-4.5 excels in factual accuracy and emotional intelligence, it struggles on complex academic benchmarks compared to leading AI models. OpenAI acknowledges the high operational costs of GPT-4.5, raising questions about its long-term availability in the API. The model is not intended to replace GPT-4o but offers enhanced features such as file uploads and improved response quality. OpenAI aims to gather insights on GPT-4.5’s capabilities through this research preview, with plans to integrate its models with reasoning approaches in future developments. Overall, GPT-4.5 represents both a significant advancement and a potential pivot point for AI training methodologies.
