OpenAI has introduced GPT-5.5, its newest flagship AI model, aiming to deliver faster reasoning, improved coding, and stronger multi-step task execution. The company announced that the model can better interpret user intent and complete complex workflows with minimal guidance. As a result, GPT-5.5 can plan tasks, use tools, verify outputs, and continue operations independently.
The rollout has already begun for Plus, Pro, Business, and Enterprise users across ChatGPT and Codex. Additionally, OpenAI confirmed that a more advanced GPT-5.5 Pro version will be available for higher-tier subscribers, while API access will follow soon.
Improved Coding Performance and Efficiency
Notably, coding remains a major focus for GPT-5.5. OpenAI reported that the model outperforms GPT-5.4 while using fewer tokens, which may reduce computing costs. On Terminal-Bench 2.0, GPT-5.5 achieved 82.7%, compared to 75.1% for its predecessor. Similarly, it scored 73.1% on the Expert-SWE benchmark, surpassing GPT-5.4’s 68.5%.
Moreover, the model handles large codebases more effectively, identifies unclear bugs, and applies changes across multiple files. These improvements make it particularly useful for software engineering and automation tasks.
Expanded Workplace and Research Capabilities
Beyond coding, GPT-5.5 enhances productivity across professional workflows. It scored 84.9% on GDPval and 78.7% on OSWorld-Verified, indicating stronger performance in office tasks and system navigation. Meanwhile, OpenAI reported increased internal adoption of Codex across departments.
In research and mathematics, GPT-5.5 also showed gains, including higher scores on GeneBench, BixBench, and FrontierMath benchmarks. Despite these improvements, the model maintains similar response speeds to GPT-5.4, supported by advanced NVIDIA infrastructure.
Overall, OpenAI emphasized stronger safety measures and confirmed expanded context limits, reinforcing its push toward more capable, autonomous AI systems.
