OpenAI Launches GPT-4.5 for ChatGPT—It’s Huge and Compute-Intensive
1 min read
Summary
OpenAI has released its new generative AI model, GPT-4.5, which is bigger and more compute-intensive than its predecessors and boasts better understanding of user prompts.
To access the research preview of GPT-4.5, users must purchase a $200 per month ChatGPT Pro subscription.
The launch follows releases from Anthropic and DeepSeek, the former’s hybrid reasoning model for its Claude chatbot and the latter’s budget-trained powerful model.
OpenAI has committed to investing billions in AI infrastructure to develop its continued focus on scaling its models, which researchers claim have captured more human emotion and interaction nuances.
While specifics on GPT-4.5’s size and compute requirements remain undisclosed, the model is incompatible with AI Voice Mode and its academic benchmark results show it was outperformed by DeepSeek’s R1 model in math but outperformed it in language benchmarks.
OpenAI’s CEO has stated that GPT-4.5 marks the end of non-chain-of-thought models, with future blends being streamlined to suit user prompts.
