← Latest news 
GPT-5.5 launches with agentic computer control, narrowly beating Claude Mythos on Terminal-Bench 2.0
Technology
Published on 24 April 2026
It wins the sandbox race, yet API access is locked
OpenAI has unveiled GPT-5.5, positioning it as a major shift toward “agentic” work that can navigate software and complete multi-step tasks with less prompting. On Terminal-Bench 2.0, it narrowly edges Anthropic’s restricted Claude Mythos Preview. But the upgrade comes with sharply higher API prices and delayed developer access, while rollout is limited to ChatGPT subscribers.
- GPT-5.5 is designed to work with minimal guidance on messy tasks
- It narrowly beats Claude Mythos Preview on Terminal-Bench 2.0
- API access is delayed and pricing rises substantially for developers
- “Trusted Access for Cyber” tightens controls for general users
Read the full story at Venture Beat
This summarization was done by Beige for a story published on 
Venture Beat