Archive — US — techcrunch
OpenAI says it’ll release o3 after all, delays GPT-5
OpenAI will release its o3 reasoning model and a new version called o4-mini shortly, despite delays in launching GPT-5. CEO Sam Altman stated that the pushback is due to efforts to enhance GPT-5's integration and capacity for expected high demand. The company aims to provide these advanced models while facing competition from more open approaches from rivals.
Key points
- OpenAI initially canceled the consumer launch of the o3 reasoning model in February but has now decided to release it together with the o4-mini in a few weeks.
- The delay in GPT-5's launch stems from the need to improve its integration and capabilities to meet high anticipated demand.
- GPT-5 will offer unlimited chat access at a standard intelligence level, while premium subscribers get higher intelligence access.
- Altman emphasizes on unifying models for broader utility, incorporating various features such as voice and deep research.
- OpenAI faces competitive pressure from models launched by rivals using an open approach to AI development.
Why it matters
- This announcement reflects trends in the AI industry towards greater competition and innovation, especially with alternatives to proprietary models.
- The increasing need for robust and unified systems in AI aligns with widespread industry interest in improving user experiences and capabilities.
- The move towards open models resonates with ongoing discussions on the ethical implications and accessibility of AI technologies.
OpenAI's strategy to release o3 and o4-mini while delaying GPT-5 highlights the company's commitment to quality and user demand amid competitive pressures, reinforcing the trend towards more robust and versatile AI models.