ChatGPT Pro Can Now Run Several Agents on One Task at the Same Time
OpenAI says Pro users can split a job across parallel agents in ChatGPT and watch their progress. Details are thin, so here is what is known and what to wait for.
OpenAI has upgraded its top ChatGPT tier so that one request can be split among several agents working at the same time. According to a post OpenAI shared on October 16, as reported by RuntimeWire, Pro users in the Chat tab can hand out parts of a task, follow each agent’s progress, and review the finished work. OpenAI also says responses are faster and that the agents use plugins and tools better.
Think of it as hiring a small team instead of one assistant. If you ask for a competitor comparison of five companies, one agent can look at each company in parallel rather than one agent doing them in a row. In theory that saves time, and each agent has a narrower job, which can help quality. A progress view lets you see which parts are finished and which are stuck.
Now the fine print, because it matters here. RuntimeWire notes that OpenAI gave no latency numbers, no examples and no measured gains in speed, quality or cost. The “better tool use” claim comes from the image caption of the post. It is also unclear how much control you have over what each agent is allowed to do with tools. So treat this as an announcement of a feature, not proof that it works well.
What’s behind it: splitting work across agents is one of the main ideas in the current wave of AI products. A single model can only hold so much in mind at once (its “context window”, roughly how much text it can keep in view), so handing out sub-tasks keeps each one focused. The trade-off is cost and coordination: more agents means more compute, and results still have to be stitched together correctly. Windows Report also warned that running many agents can strain your computer’s processor, though that concerned a separate setup and is not confirmed for this feature.
What this means for you: if you are on a free or Plus plan, nothing changes. If you are a Pro user, try it on a job that naturally divides into parts, such as research across several sources or checking several documents, and compare with one agent doing it. Read each agent’s output before you rely on it, especially when tools can act on your behalf. For everyone else, a good “wait and see” feature.
Sources
Source: https://runtimewire.com/article/openai-gpt-6-pro-chat-multi-agent-upgrade
Google's Harness²: An Agent That Improves Its Own Toolbox Instead of Its Brain
An open-source research project from Google Research lets an agent rewrite its own prompts, tools and memory for each task while the model stays frozen. Gains are real but the safety defaults are loose.
