OpenAI’s GPT-5.5 has posted an 82.7% score on Terminal-Bench 2.0, a benchmark that throws AI agents into difficult, real-world command-line tasks and grades them with zero partial credit. According to results circulating among AI researchers this month,

Tencent just open-sourced Hy3 preview, a model that punches above its weight on coding agents, reasoning, and search—built in under 3 months.

OpenAI Releases GPT-5.5, a Fully Retrained Agentic Model That Scores 82.7% on Terminal-Bench 2.0 and 84.9% on GDPval