Archive — US — techcrunch
OpenAI releases GPT-5.5 Instant, a new default model for ChatGPT
OpenAI has launched GPT-5.5 Instant as the new default model for ChatGPT, boasting reduced hallucination in key areas and enhanced context management. The model achieved impressive AIME 2025 and MMMU-Pro scores, improving coding and knowledge tasks. Features include personalized answers and memory source visibility. The release response follows past user backlash against GPT model withdrawals.
Key points
- OpenAI released GPT-5.5 Instant, replacing GPT-5.3 Instant as the default ChatGPT model.
- The new model reduces hallucinations in law, medicine, and finance, while maintaining low latency.
- GPT-5.5 scored 81.2 on the AIME 2025 math test, compared to 65.4 for GPT-5.3.
- It outperformed GPT-5.3 on the MMMU-Pro multimodal reasoning benchmark with a score of 76 vs. 69.2.
- Enhanced context management allows the model to refer back to past conversations and files for personalized responses.
- Memory source visibility will be implemented for Plus and Pro users with plans for broader access.
- Developers can access GPT-5.5 via API as 'chat-latest', with GPT-5.3 remaining for paid users for three months.
- OpenAI faces past backlash over model withdrawals, as seen with GPT-4o's retirement despite user popularity.
Why it matters
- The rapid evolution of AI models in 2025 reflects a broader trend towards personalized technology in communication and work.
- Concerns over AI reliability in critical sectors underline the importance of accuracy in AI models following incidents in 2024.
- The user backlash against model withdrawals highlights the emotional attachment users develop towards AI technologies.
The launch of GPT-5.5 Instant signifies OpenAI's commitment to improving AI reliability and user experience, while navigating user sentiments towards previous model discontinuations.