OpenAI to Prioritize Smaller Models; GPT-5 Delay Announced
20 April 2025 · Uncategorized ·
Source: · https://technews.tw/2025/04/05/openai-is-going-to-release-o3-and-o4-mini-after-all/
OpenAI has revised its product roadmap, announcing that it will release the standalone models O3 and O4-mini before postponing GPT-5. Initially slated for a mid-February simplification of their plans with no immediate O3 release but instead previewing it alongside GPT-5 later on, this change was revealed by CEO Sam Altman.
“OpenAI is still planning to introduce both O3 and O4-mini in the coming weeks, followed by GPT-5 several months later,” Altman stated on X (formerly Twitter). He attributed the shift to challenges encountered during the development of GPT-5, which will now be released after its initial schedule.
Altman explained that this decision stemmed from multiple factors, including an opportunity to significantly enhance GPT-5’s performance beyond original expectations and difficulties in seamlessly integrating all functionalities while ensuring sufficient resources for anticipated demand. "We are going to be able to make GPT-5 much better than originally..." he wrote on X (formerly Twitter).
Previously, Altman had indicated that OpenAI intends to deploy GPT-5 without restrictions under standard intelligence settings but will prohibit misuse. Furthermore, paying users of ChatGPT Plus and Pro versions would gain access to higher levels of model intelligence integrating voice capabilities, Canvas, search functionality, Deep Research tools, and other features exclusive for paid subscribers.
OpenAI’s closed strategy contrasts with competitors like DeepSeek, who are adopting an “open” approach by releasing models to AI communities and commercializing them under certain conditions. The recent launch of ChatGPT's image generation capabilities has also contributed to GPU shortages; OpenAI is actively working on enhancing computational resources and efficiency to meet the growing demand.
“OpenAI is still planning to introduce both O3 and O4-mini in the coming weeks, followed by GPT-5 several months later,” Altman stated on X (formerly Twitter). He attributed the shift to challenges encountered during the development of GPT-5, which will now be released after its initial schedule.
Altman explained that this decision stemmed from multiple factors, including an opportunity to significantly enhance GPT-5’s performance beyond original expectations and difficulties in seamlessly integrating all functionalities while ensuring sufficient resources for anticipated demand. "We are going to be able to make GPT-5 much better than originally..." he wrote on X (formerly Twitter).
Previously, Altman had indicated that OpenAI intends to deploy GPT-5 without restrictions under standard intelligence settings but will prohibit misuse. Furthermore, paying users of ChatGPT Plus and Pro versions would gain access to higher levels of model intelligence integrating voice capabilities, Canvas, search functionality, Deep Research tools, and other features exclusive for paid subscribers.
OpenAI’s closed strategy contrasts with competitors like DeepSeek, who are adopting an “open” approach by releasing models to AI communities and commercializing them under certain conditions. The recent launch of ChatGPT's image generation capabilities has also contributed to GPU shortages; OpenAI is actively working on enhancing computational resources and efficiency to meet the growing demand.