OpenAI Rolls Out GPT-4.5 with Enhanced Creativity and Lower Hallucination Rate
OpenAI has introduced GPT-4.5, its latest AI model, in a research preview for ChatGPT Pro users and developers worldwide. The phased rollout will extend in the coming weeks, with Plus and Team users gaining access next week, followed by Enterprise and Edu users.
Key Enhancements in GPT-4.5:
- Improved Pattern Recognition & Creativity
- GPT-4.5 enhances pattern identification and generates more creative insights, though its reasoning abilities remain comparable to earlier models.
- Greater Emotional Intelligence
- The model better understands human emotions, leading to more natural interactions in text-based conversations.
- Lower Hallucination Rate
- False information generation has dropped to 37.1%, a significant improvement from GPT-4o’s 61.8% and the older reasoning model’s 44%.
- Expanded Capabilities
- GPT-4.5 supports file and image uploads, assisting with writing and coding projects, though voice and video features are yet to be included.
CEO Sam Altman called GPT-4.5 a “giant, expensive model” and cited GPU shortages as a challenge in the gradual rollout. OpenAI, backed by Microsoft, is adding tens of thousands of GPUs to meet demand and further refine its AI capabilities.
