OpenAI presents GPT-5.5 with new autonomous capabilities and focus on more intuitive computing

May 6, 2026

TL;DR

OpenAI launched GPT-5.5, an AI model with increased autonomy and user intent understanding.
The update is being implemented in ChatGPT plans and will be available via API soon.
GPT-5.5 is designed to handle multi-step processes with less human intervention and better autonomous planning.
Key improvements focus on autonomous programming, computer usage, scientific research, and specialized knowledge tasks.
The model can manage ambiguous processes, adapt to sustained contextual reasoning, and perform tasks like code writing/debugging, online research, and data analysis.
GPT-5.5 maintains similar speeds to GPT-5.4 and uses fewer tokens for equivalent tasks in Codex.
Internal tests show GPT-5.5 achieving 84.9% on GDPval and 78.8% on OSWorld-Verified.
New integrated versions include GPT-5.5 Thinking (for programming and complex analysis) and GPT-5.5 Pro (for business users with enhanced accuracy and structure).
Security measures have been enhanced, incorporating OpenAI's 'most robust protection system to date'.
GPT-5.5 Pro offers more complete, structured, and precise answers for business, legal, educational, and data science applications.

Continue reading the original article