Manus unveils new Wide Research mode, enabling 100+ parallel AI agents for scalable tasks

Chinese AI startup Manus has launched Wide Research, a feature that spins up over 100 parallel AI agents to handle large-scale tasks simultaneously. Unlike Deep Research tools from other AI models from like OpenAI's ChatGPT, Mistral Le Chat or Google Gemini that process tasks sequentially, Wide Research takes another turn and emphasizes speed and broader coverage by distributing tasks across many agents at once.

Each session runs in its own virtual machine, allowing users to execute complex workflows—like generating presentations or testing open-source tools by simply interacting with the agent. These agents function as fully Turing-complete cloud environments, removing the need for technical setups. In one demo, it deployed 100 subagents to analyze sneaker designs, pricing, and availability, producing sortable outputs in spreadsheet and web formats. Another test generated 50 unique poster designs in parallel, returned as a ZIP file. Unlike fixed-role multi-agent systems, each subagent is a fully capable Manus instance with general-purpose capabilities, supported by a new virtualization architecture that reportedly scales compute power 100x.

The feature activates automatically when appropriate and is currently available only to Pro plan users ($199/month), with a gradual rollout planned for Plus and Basic tiers. Despite promising infrastructure, Manus hasn’t shared benchmarks comparing Wide Research’s effectiveness or efficiency to conventional methods.