OpenAI Unveils Performance-Boosted 'GPT-5.5', Intensifying Competition

Artificial intelligence (AI) developer OpenAI has entered the race for market leadership by releasing a new AI model, 'GPT-5.5', with improved performance and autonomy. Unveiled on the 23rd of this month (local time), GPT-5.5 is described as the smartest and most intent-aware AI to date, drawing attention in its technological competition against rivals like Anthropic's 'Claude Opus 4.7'.

Benchmark reports indicate that GPT-5.5 surpassed Claude Opus 4.7 in knowledge work and cybersecurity. Specifically, in an evaluation of terminal environment task capabilities, GPT-5.5 achieved 82.7%, exceeding Opus 4.7's 69.4% by over 10 percentage points. However, in the 'SWE-Bench PRO' metric for coding, GPT-5.5 scored 58.6%, lagging behind Opus 4.7's 64.3% by more than 5 percentage points.

Greg Brockman, President of OpenAI, highlighted 'autonomy' – the ability to grasp tasks with minimal instructions – as a core feature of GPT-5.5, emphasizing that it will form the foundation for the next era of computation.