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

MarkTechPost / 4/24/2026

📰 NewsIndustry & Market MovesModels & Research

共有:

Key Points

OpenAI has released GPT-5.5, an agentic model designed to handle the full computer-work stack, including coding, research, data analysis, and software operation.
The model is positioned to reduce the need for human supervision by enabling more autonomous end-to-end task execution.
GPT-5.5 reported strong benchmark results, scoring 82.7% on Terminal-Bench 2.0.
It also achieved 84.9% on GDPval, reinforcing its performance across evaluation setups.
Overall, the release emphasizes fully retrained capabilities aimed at reliable agentic workflows rather than step-by-step human guidance.

The model targets the full stack of computer work — coding, research, data analysis, and software operation — without needing a human to supervise every step

The post OpenAI Releases GPT-5.5, a Fully Retrained Agentic Model That Scores 82.7% on Terminal-Bench 2.0 and 84.9% on GDPval appeared first on MarkTechPost.