FrontierThe story, in brief
OpenAI Releases GPT-5.5, a Fully Retrained Agentic Model That Scores 82.7% on Terminal-Bench 2.0 and 84.9% on GDPval
82.7% on Terminal-Bench. OpenAI's GPT-5.5 isn't just smarter—it's autonomous across your entire stack.
The KeyNews take
Why it matters
OpenAI released a fully retrained agentic model targeting autonomous execution across coding, research, data analysis, and software operations—a significant capability leap that redefines what 'autonomous' means in production AI workloads.
The key facts
6 to knowGPT-5.5 fully retrained with agentic capabilities
82.7% on Terminal-Bench 2.0
84.9% on GDPval benchmark
Targets full-stack computer work: coding, research, data analysis, software operation
Operates without requiring human supervision per step
Published April 23, 2026 on MarkTechPost
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: The model targets the full stack of computer work — coding, research, data analysis, and software operation — without needing a human to supervise every step