GPT-6 Astra Lands in ChatGPT Work and Codex, OpenAI Claims Edge Over Claude on Coding
OpenAI extended GPT-6 Astra into ChatGPT Work and Codex, saying it scores 57.9% on its internal coding benchmark versus 55.8% for Claude Fable 5.1, while also operating inside business applications that lack APIs.
why it mattersA direct, self-reported benchmark shot at Anthropic's flagship model sharpens the competition for enterprise coding and agent work.