OpenAI has launched GPT-5.4, introducing specialized Pro and Thinking versions optimized for professional work. The new model features a massive 1 million token context window and shows significant improvements in token efficiency and benchmark performance. It demonstrates 33% fewer factual errors compared to GPT-5.2 and excels at complex tasks like legal analysis and financial modeling.
Background
OpenAI has been progressively developing more advanced GPT models, with each iteration improving capabilities while addressing limitations like hallucinations and context limitations. The AI industry has been competing to develop models with longer context windows and better reasoning abilities.
- Source
- TechCrunch
- Published
- Mar 6, 2026 at 02:00 AM
- Score
- 8.0 / 10