OpenAI Launches Multi-Agent Parallel Execution for GPT-6 Astra Pro in ChatGPT
OpenAI has updated GPT-6 Astra Pro in ChatGPT with multi-agent parallel execution and faster tool handling, available exclusively to Pro users in the Chat tab.
Image source: https://eu.36kr.com/en/p/4031321049715076
On October 17, 2026, OpenAI announced an upgrade to GPT-6 Astra Pro in ChatGPT, adding multi-agent parallel execution to handle different parts of a task simultaneously. This structural shift moves beyond sequential generation, improving response speed and introducing a modular workflow for complex conversational requests.
Parallel Task Delegation and Subtask Tracking
OpenAI confirmed that multiple agents can now divide a single objective and work in parallel. Rather than waiting for a single thread to complete multi-step instructions sequentially, users see subtasks distributed across concurrent agents. The interface provides individual progress tracking for each agent in real time. Once an agent finishes its assignment, users can immediately open the completed task card to review intermediate findings, code snippets, or structured data without waiting for the broader prompt to wrap up. This granular visibility helps verify outputs early and ensures transparent execution throughout complex tasks.
Tool Optimization and Workflow Impact
In addition to parallel agents, OpenAI upgraded GPT-6 Astra Pro's response latency and enhanced its integration with plugins and tools. By allowing separate agents to invoke external capabilities at the same time—such as querying documentation, querying APIs, or parsing structured files—the system reduces overall wall-clock time for composite workflows. For technical users, parallel tool invocation removes single-thread queuing bottlenecks and keeps each external action isolated within its designated subtask, making multi-step problem solving more manageable and easier to evaluate.
Subscription Tiers, Usage Caps, and Caveats
This capability includes notable access restrictions. OpenAI specified that parallel multi-agent processing is available exclusively to Pro tier subscribers in the Chat tab, leaving Free and Plus accounts without access. Early community feedback notes that geographic rollout timelines may vary, and monthly Pro message quotas apply when assigning intensive jobs to multiple agents. Furthermore, some users observed that the change represents an ergonomic workflow upgrade rather than a fundamental leap in underlying intelligence, meaning teams must assess whether their tasks truly require parallel agent coordination.