OpenAI has released its latest and largest AI model yet — GPT-4.5. According to OpenAI, the new model has “deeper world knowledge, with a better understanding of user intent,” and has been designed for “broad general knowledge, including writing help, coaching, brainstorming, and nuanced communication.”
GPT-4.5 supports function calling, Structured Outputs, vision, streaming, system messages, evals, and prompt caching. It’s currently available via OpenAI’s Completions, Assistants, and Batch APIs.
READ: AI and the media: Transforming content, distribution, and editorial leadership (December 10, 2024)
A typical query costs on average $68 per 1 million tokens with cache discounts ($75 / 1M input tokens, $37.5 /1M cached input, $150 / 1M output). Batch jobs are discounted 50% and cached input is discounted 50%.
OpenAI also said “GPT-4.5 builds on GPT-4o to be a more general-purpose model rather than focusing primarily on STEM-related reasoning. The model has been trained using a combination of new supervision techniques and traditional methods like supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF), similar to GPT-4o.”
The AI juggernaut also mentioned that GPT-4.5 has been trained on data derived from smaller models, enhancing its steerability and ability to grasp nuance. This makes interactions seem more natural, and ensure there is better alignment with user intent.
OpenAI also claims the model doesn’t “think before responding,” rather it excels at interpreting subtle cues and demonstrating emotional intelligence, making it well-suited for tasks such as writing, designing, programming, and solving practical problems. In terms of safety, the company noted that it has found no significant increase in risks when compared to previous models.
READ: Regulating AI: Sanjay Puri on policy, challenges, and ethical innovation (November 1, 2024)
GPT-4.5 is currently available for ChatGPT Pro subscribers on web, mobile, and desktop through the model picker. OpenAI will roll it out to Plus and Team users next week, followed by Enterprise and Edu users the following week.
This comes at a time of stiff competition in the AI market. Chinese startup DeepSeek had made great waves in the industry with its release of its AI model noted to be more efficient and cost-effective than other AI models. Elon Musk’s xAI had also recently unveiled its new AI model Grok 3. Anthropic also recently released its first “hybrid reasoning” model.