Quick Buzz Feed
The model is better at coding, using computers and pursuing deeper research capabilities, OpenAI says.
OpenAI has officially launched GPT-5.5, its newest artificial intelligence model, demonstrating enhanced capabilities in critical areas such as coding, efficient computer utilization, and advanced research. This announcement comes just a few weeks after a significant release from competitor Anthropic, which unveiled its Claude Mythos Preview model, known for its sophisticated cybersecurity features. GPT-5.5 is being progressively rolled out to OpenAI's various paid subscriber tiers, specifically including Plus, Pro, Business, and Enterprise users, and is accessible through both ChatGPT and its specialized coding assistant, Codex.
GPT-5.5 represents a notable advancement in artificial intelligence, with OpenAI emphasizing its ability to perform with 'less guidance,' allowing it to interpret complex problems and autonomously determine necessary next steps. OpenAI President Greg Brockman highlighted this aspect, suggesting it lays a new foundation for how users will interact with computers and execute tasks. The rapid introduction of GPT-5.5, shortly after the previous GPT 5.4 release, highlights the intense and fast-paced competitive environment within the AI sector, where companies like Google and Anthropic are also aggressively developing and releasing advanced models like Claude Mythos Preview.
OpenAI's GPT-5.5 exhibits strong proficiencies across a range of tasks, including detailed data analysis, proficient code writing and debugging, seamless software operation, comprehensive online research, and efficient creation of documents and spreadsheets. Addressing potential concerns, OpenAI noted that while the model meets its 'High' risk classification for cybersecurity—indicating it could 'amplify existing pathways to severe harm'—it does not exceed the 'Critical' threshold, which implies 'unprecedented new pathways to severe harm.' Mia Glaese, OpenAI's vice president of research, confirmed that GPT-5.5 underwent rigorous third-party safeguarding and 'red teaming' for both cyber and biological risks, with continuous improvements to its security protocols. This careful approach is particularly relevant following Anthropic's decision to limit the rollout of its Mythos model due to its capacity for identifying software vulnerabilities. The deployment of GPT-5.5 to OpenAI’s paid subscribers (Plus, Pro, Business, and Enterprise users) in ChatGPT and Codex is underway, with its application programming interface (API) release slated 'very soon,' requiring distinct safeguards.