OpenAI has launched GPT-4.5 under the Orion name to become its most advanced AI model yet. The newest version represents the company’s largest model, which received more extensive training using enhanced computing resources compared to previous versions. OpenAI first declared through a white paper that GPT-4.5 does not belong to the category of “frontier AI model” until the organization removed this statement from the document within hours of its release.
The GPT-4.5 model is available to users who subscribe to ChatGPT Pro at $200 per month. OpenAI’s API developers and users of ChatGPT Plus and ChatGPT Team will receive access to the model during the upcoming week. The industry pays close attention to this launch because it demonstrates traditional AI training methods, which require increased computational power and additional data.
Previous generations experienced major advancements in mathematics writing and coding when model size increased. According to OpenAI, GPT-4.5 offers better world knowledge and stronger emotional intelligence than previous versions along this same pattern. However, benchmarks suggest diminishing returns. The current factual accuracy and creative capabilities of GPT-4.5 prove successful, yet it performs less competently than modern competitor AI reasoning systems such as DeepSeek and Anthropic.
OpenAI has revealed that GPT-4.5’s operating expenses are too expensive, which creates doubts about its continuous availability through the API. The payment structure for developers who use GPT-4.5 requires them to pay $75 for each million input tokens and $150 for each million output tokens, which surpasses the pricing of GPT-4o from OpenAI.
The limitations of GPT-4.5 lead experts to believe it represents a development toward OpenAI’s future combination of its “O” reasoning series with GPT-5, which will launch this year. The Orion model does not represent the ultimate AI leadership, but it creates the foundation for better and more efficient AI systems that will emerge in the future.