OpenAI added GPT-6 Astra and GPT-6.1 Terra to ChatGPT Dots, which can choose among models or combine them for a task. *(Nov. 3)
On Nov. 3, OpenAI said ChatGPT Dots would draw on GPT-6 Astra and GPT-6.1 Terra, expanding beyond Astra, which powered the agents at launch. A Dot can choose a model for each task and switch models between stages, according to OpenAI staffer Thibault Sottiaux; the company has not explained how selection works or whether users can override it. The change aims to pair demanding work with Astra and faster, cheaper tasks with Terra, while hiding model choice from users. That could make capable agents less costly to run and intensify pressure on rivals to offer model-routing agents rather than a single chatbot model. Terra launched in October for ChatGPT Work and Codex users and through the API at $1 per million input tokens and $5 per million output tokens—one-tenth Astra’s standard token prices. OpenAI says Terra matches Astra on a computer-use benchmark at that price and completes tasks more than three times faster on average. Dots, introduced Sept.