Back to AIBriefs
LaunchAI AgentsDevelopers

Webwright: Microsoft's terminal-native web agent scores 60.1% on Odysseys

Webwright: Microsoft's terminal-native web agent scores 60.1% on OdysseysMarkTechPost

Webwright achieved 60.1% on the Odysseys benchmark, nearly doubling base GPT-5.4's 33.5%. The framework lets agents write Playwright code in a terminal environment, treating browser automation as iterative scripting instead of sequential actions.

·
17 days ago
Webwright: Microsoft's terminal-native web agent scores 60.1% on Odysseys — AIBriefs