OpenAI releases GPT-5.5 with agentic benchmarks, 20% price hike, and 'super app' vision

OpenAI released GPT-5.5, described as its 'smartest and most intuitive' model yet, scoring 82.7% on Terminal-Bench 2.0 and 73.1% on Expert-SWE. The model costs 20% more via API than its predecessor but delivers lower token usage at matching latency. CEO Sam Altman framed the release as 'iterative deployment' central to OpenAI's safety strategy, while President Greg Brockman called it 'a new class of intelligence.' Despite benchmark gains, the model still exhibits frequent hallucinations, and chief scientist Jakub Pachocki acknowledged AI progress has been 'surprisingly slow' but promised 'extremely significant improvements' ahead.