AI models are getting very good at professional tasks, new OpenAI research shows

By admin
Anthropic’s Claude Opus 4.1 was especially good at tasks performed by clerks, software developers, and private investigators