OpenAI launched GPT-4.5, Most Expensive and Knowledgeable GPT Model Yet

OpenAI launched ChatGPT-4.5, the latest, largest, most knowledgeable, and most expensive GPT model yet. The AI tech company released the GPT-4.5 model as a research preview.

This model enhances upon previous iterations by improving writing capabilities, expanding world knowledge, and refining its interactive personality, resulting in more natural and intuitive user experiences.

Notably, GPT- 4.5 demonstrates a significant reduction in hallucinations, with rates decreasing from nearly 60% in GPT-4o to 37% in GPT-4.5. This is “OpenAI’s largest LLM, improving on GPT-4’s computational efficiency by more than 10x.” 

At its initial stage of release, only ChatGPT Pro users can access it in the ChatGPT interface. The rest of the paid users of ChatGPT will be able to use it in the coming weeks.

What is GPT 4.5?

According to OpenAI, ChatGPT 4.5 is “generally useful and inherently smarter…. It’s the best chat model yet.” This model has deeper knowledge and better contextual understanding than the previous GPT models.

Unlike the o-series reasoning models of OpenAI, GPT 4.5 is an unsupervised learning model which lets it improve its accuracy by reducing the hallucination rate. GPT-4.5 is an effort to scale up unsupervised learning, and resultantly, acquire world knowledge intuition and reduce hallucination.

Use our Chat GPT free chatbot for unlimited use without registration.

The accuracy rate of GPT-4.5 is 62.5% which is almost double the GPT-4o accuracy level of 38.2%. Even the OpenAI’s reasoning models like o1 and o3-mini score less in this regard compared to GPT-4.5 with 47% and 15% accuracy rates respectively.

Since ChatGPT 4.5 scores higher in terms of its accurate responses, it has a much lower hallucination rate as compared to previous GPT and reasoning models.

GPT- 4.5 has improved writing capabilities, deeper world knowledge, and a “refined personality over previous models.” It can be quite effective for practical tasks like writing, learning, coaching, communication, and brainstorming. etc. It’s more of a collaborative model – “more intuitive and emotionally nuanced.”

OpenAI states that interactions with GPT- 4.5 would be more natural and fluid, because of its enhanced ability to recognize patterns, identify relationships, and draw meaningful connections. These advancements make GPT- 4.5 particularly well-suited for “writing, programming, and practical problems” with greater accuracy and efficiency.

Try ChatGPT 4 free of cost with web search capability to get latest data.

Who Can Access ChatGPT 4.5?

ChatGPT Pro users and the developers (all paid tiers) in the API can access ChatGPT 4.5. The company intends to release it to ChatGPT Plus, Team, and Edu users in the next week. Whereas ChatGPT Enterprise users could use the model in the coming following week.

How Much Does GPT-4.5 Cost in API?

The latest GPT model by OpenAI, the ChatGPT owner company, is the largest GPT model, and it takes tons of compute to operate fast and efficiently which makes it super expensive. ChatGPT-4.5 model’s input cost is 2900% higher than GPT-4o, and its output cost is 1400% higher.

Similarly, it costs much more than the company’s latest reasoning model. o3-mini’s input cost is 98.53% lower than GPT- 4.5, and its output cost is 97.07% lower.

Here are the prices of GPT- 4.5 and GPT- 4o:

Below is the cost comparison of GPT 4.5 and the latest reasoning model o3-mini.

Final Thoughts

While GPT-4.5 is OpenAI’s largest and most knowledgeable AI model and produces more accurate responses, it’s the most expensive model available. With OpenAI releasing more advanced AI models, the cost of the latest AI is increasing making it inaccessible to free ChatGPT users.

I believe quality AI should be affordable and thus accessible to everyone. Tech giants should make efforts to make advanced AI in the reach of everyone.