OpenAI upgraded GPT-6 Pro in ChatGPT on October 16, 2026: faster responses, better use of plugins and tools, and several agents that work on parts of one task at the same time. The ChatGPT release notes list “Multiple agents that can work on different parts of a task in parallel.” You can follow their progress in the conversation and open each completed task to see what that agent did.
OpenAI announced it in a post on X at 20:00 UTC: “GPT-6 Pro in ChatGPT got an upgrade with faster responses and improved use of plugins and tools.” The update is rolling out to eligible Pro users in the Chat tab.
OpenAI upgraded GPT-6 Pro in ChatGPT on October 16, 2026, and listed it in the ChatGPT release notes.
Multiple agents can now work on different parts of one task in parallel, and you can open each completed task to see what that agent did.
The same update brings faster responses and improved use of plugins and tools.
GPT-6 Pro is powered by GPT-6 Astra and available on Pro $100, Pro $200, Business and Enterprise plans; Plus does not get it in Chat.
OpenAI has not said how many agents a task can use, whether they share context, or how parallel runs count against limits.
Anthropic’s Claude Code Projects and CellCog subagents also run parallel work; CellCog’s helpers run under an AI employee that keeps memory between sessions.
§ 01What changed
| Change | OpenAI’s description | What it means for you |
|---|---|---|
| Speed | Faster responses | Pro answers arrive sooner |
| Tools | Improved use of plugins and tools to carry out tasks | Fewer stalls when Pro calls connected apps |
| Parallel agents | Multiple agents work on different parts of a task in parallel | One request can fan out into several workers |
| Visibility | Follow progress in the conversation, open completed tasks | You can check each agent’s work, not just the final answer |
| Rollout | Eligible Pro users; managed workspaces depend on settings | Business and Enterprise access is not guaranteed yet |
GPT-6 Pro is the top effort level in ChatGPT’s model menu. OpenAI’s models help article says it is “powered by GPT-6 Astra” and available on the Pro $100, Pro $200, Business and Enterprise plans. Plus subscribers get Astra in ChatGPT Work and Codex but not GPT-6 Pro in Chat. The same article notes that Intelligent UI, GPT-6’s visual answers, is not available at Pro effort.
§ 02What OpenAI has not said yet
The release note is four lines long. It leaves out the questions a team would ask before relying on the feature:
- How many agents a task can use, and whether you choose the split or the model does.
- Whether the agents share context with each other, or each sees only its own piece.
- Limits. Nothing on whether parallel runs draw faster on the Pro usage allowance.
- Whether work continues after you leave the conversation.
- API access. The OpenAI API changelog for October 16 lists voice updates only.
One early reply to OpenAI’s post said Pro agents sometimes report the GitHub plugin as read-only. That is a single user’s report, not a documented limit.
§ 03How it compares
Parallel agents are now a standard move. Anthropic shipped Claude Code Projects on October 10, where one conversation directs parallel cloud threads with shared memory, in beta for select Pro and Max users.
| Product | Who gets it | How the work splits | How you check it |
|---|---|---|---|
| ChatGPT GPT-6 Pro | Eligible Pro users, rolling out | Multiple agents on parts of one task | Follow progress, open completed tasks |
| Claude Code Projects | Select Pro and Max users, beta | A coordinator runs parallel cloud threads with shared memory | The coordinating conversation |
| CellCog | Every account | An AI employee or chat starts helper agents, or hands work to other AI employees | The employee reviews each result before it calls the job done |
The difference is what persists. ChatGPT’s agents live inside one Pro conversation. CellCog’s subagents run under an AI employee that keeps its own computer, task board and memory between sessions, so the next request starts from the last result instead of from a blank chat.
§ 04Should you switch on Pro for this
If you already pay for Pro, try it on a task with separable parts, such as researching five vendors at once or drafting sections of a report. Open each completed task and check it; OpenAI built the view for exactly that. If you are on Plus, this update alone does not change your plan: GPT-6 Pro still requires Pro, Business or Enterprise.
Q1Do I need ChatGPT Pro for parallel agents?
Yes, for now. The release notes say the update is rolling out to eligible Pro users; GPT-6 Pro is also on Business and Enterprise plans, subject to workspace settings.
Q2Can I see what each agent did?
Yes. OpenAI says you can follow their progress in the conversation and open completed tasks to see what each agent did.
Q3What model powers GPT-6 Pro?
GPT-6 Astra, according to OpenAI’s help article on GPT-6 and other models in ChatGPT.
Q4Does Intelligent UI work with GPT-6 Pro?
No. OpenAI’s help article says Intelligent UI is not available at Pro effort.
Q5What model does CellCog run?
Every CellCog tier runs Claude Opus 5.5.