On October 16, 2026, OpenAI announced that GPT-6 Pro in ChatGPT can assign different parts of a task to multiple agents working in parallel. Users can follow their progress in the conversation and open completed tasks to inspect each agent’s work. OpenAI described the rollout as gradual for eligible Pro users; access in managed workspaces depends on workspace settings and rollout eligibility. OpenAI’s release notes
What GPT-6 Pro’s parallel agents do
The feature splits work across different parts of a task, rather than presenting the user with a single stream of progress. In Chat, users can follow that progress in the conversation and inspect completed tasks. That visibility gives you a way to review the agents’ contributions as the task moves forward.
OpenAI’s model guide lists GPT-6 Pro for Pro $100, Pro $200, Business, and Enterprise plans. Enterprise access also depends on workspace model permissions. The guide says GPT-6 Pro is powered by GPT-6 Astra, the model covered in NeoTeo’s earlier report.
The allowance rule for Business plans
Business Standard and Business Premium share their included allowance across GPT-6 Pro and GPT-5.6 Sol Pro, according to OpenAI’s model guide. Choosing between those two models therefore draws on the same included allowance for these plans.
How Chat’s agents relate to Ultra
OpenAI describes Ultra in Work or Codex as a reasoning mode that uses subagents to handle separate parts of complex work in parallel. The October 16 announcement concerns GPT-6 Pro in the Chat tab. Both descriptions involve parallel work, but they refer to different ChatGPT surfaces. OpenAI’s models documentation
