August 4, 2025
5 min read
Markus Kasanmascheff
Manus AI unveils Wide Research, deploying 100+ AI agents in parallel to revolutionize large-scale research and challenge single-agent models.
Manus AI, a Singapore-based startup, has unveiled "Wide Research," a significant new feature for its agentic platform. This innovative capability allows users to deploy over 100 AI agents in parallel to tackle large-scale research tasks. Announced on July 31, 2025, Wide Research distinguishes itself from competitors like OpenAI and Google by focusing on broad parallel processing rather than "deep research."
Manus positions Wide Research as a personal supercomputing cluster, enabling users to manage complex, high-volume research through natural language prompts. The feature is initially available to Manus Pro subscribers, with a phased rollout planned for other subscription tiers. This launch signifies a new direction for scaling AI-driven productivity.
The development of Wide Research builds upon the considerable interest Manus garnered in March 2025. Initially thought to be a Chinese AI venture, its parent company, Butterfly Effect, is now headquartered in Singapore, a strategic move in response to US-China tech tensions. The platform's initial launch saw overwhelming demand, with partner Zhang Tao acknowledging they had "completely underestimated the level of enthusiasm."
Manus describes its fundamental offering not as an AI, but as a "personal cloud computing platform." The company's vision, as articulated in a recent blog post, is to "democratize that power." Each Manus session runs on a dedicated cloud-based virtual machine, allowing users to orchestrate complex cloud workloads simply by interacting with an agent. The core architecture, where each agent functions as a Turing-complete virtual machine, underpins the new Wide Research feature.
Source: Manus AI Launches ‘Wide Research,’ Pitting 100-Agent Swarms Against ‘Deep Research‘ from Google and OpenAI
How ‘Wide Research’ Reimagines AI Collaboration
Wide Research is fundamentally designed for parallel processing and agent-to-agent collaboration. Instead of a single agent delving deeply into a task, Manus deploys a swarm of agents that work in concert. This effectively provides users with the power of a personal supercomputing cluster, all accessible through conversational commands. The system aims to amplify the available compute for a single task by a factor of 100, thereby transforming the approach to complex problem-solving. Practical applications showcase the breadth of this new feature. In a demonstration video, Wide Research was tasked with comparing 100 different sneakers. It instantaneously deployed 100 concurrent subagents, each analyzing a single shoe. The result was a sortable matrix presented in both spreadsheet and webpage formats within minutes. Another example highlighted its creative capabilities, with agents simultaneously generating 50 distinct poster designs and delivering them in a single ZIP file. Manus emphasizes that "The key to Wide Research isn’t just having more agents — it’s how they collaborate." This collaborative architecture is a critical differentiator. Unlike traditional multi-agent systems that assign predefined, specialized roles, every subagent in Wide Research is a fully capable, general-purpose Manus instance running on its own virtual machine. This generality is crucial for the system's flexibility, ensuring that tasks are not limited by rigid formats or domains.A Promising Architecture with Unproven Benefits
Wide Research is currently accessible to users on the $199/month Pro plan, with plans to roll it out to Plus ($39/month) and Basic ($19/month) users subsequently. However, the company has characterized the feature as experimental, and while the claims are ambitious, its practical benefits are yet to be fully substantiated. This cautious rollout reflects both the novelty of the architecture and the significant technical and regulatory challenges the platform may encounter. Industry observers have noted the lack of empirical data supporting the "wide" approach. As reported by VentureBeat, "while the feature showcases architectural ambition, its practical benefits over simpler methods remain unproven based on the information provided." Manus has not yet released performance benchmarks, direct comparisons to single-agent alternatives, or detailed technical explanations of how subagents collaborate, results are merged, or if the system offers demonstrable advantages in speed, accuracy, or cost-efficiency. These are not minor considerations. The broader tech community has a checkered history with multi-agent systems, which are notoriously difficult to develop and debug. Developers on other platforms have reported issues such as high token consumption, slow performance, and a lack of execution visibility. Early independent reviews of Manus itself indicated performance inconsistencies, including the generation of simulated data or agents becoming stuck in loops. These challenges underscore the significant technical hurdles Manus faces in ensuring its agent swarms are both reliable and efficient. Furthermore, the company's focus on autonomous operation has drawn considerable regulatory attention since its debut. Shortly after its initial launch, Tennessee Governor Bill Lee banned Manus from state networks, citing risks of "censorship, propaganda, and bias." Alabama soon followed with a similar prohibition due to security vulnerabilities. This governmental pushback reflects a global apprehension towards fully autonomous systems that operate without human oversight, a principle core to the Manus platform's design. Ultimately, Wide Research represents a bold stride in scaling AI. As the agentic AI market continues to mature, the distinction between ambitious vision and practical utility will become increasingly clear. The success of Wide Research hinges on its ability to overcome the well-documented challenges of multi-agent coordination and navigate a skeptical regulatory landscape to deliver tangible benefits over more established, simpler methods.Frequently Asked Questions (FAQ)
Q: What is Manus AI's Wide Research feature? A: Wide Research is a new feature on Manus AI's agentic platform that allows users to deploy over 100 AI agents simultaneously to tackle large-scale research tasks, functioning like a personal supercomputing cluster. Q: How does Wide Research differ from competitors like OpenAI? A: Unlike competitors focusing on "deep research" with single, powerful agents, Wide Research emphasizes parallel processing by deploying numerous agents concurrently to address tasks. Q: What are the practical applications of Wide Research? A: Examples include comparing numerous products (like 100 sneakers) or generating multiple creative outputs (like 50 poster designs) simultaneously, delivering results efficiently. Q: What is the underlying architecture that enables Wide Research? A: The feature is built on Manus's core architecture where each agent is a Turing-complete virtual machine, allowing for parallel processing and collaboration among a swarm of agents. Q: Is Wide Research available for all Manus users? A: Currently, Wide Research is available to Manus Pro subscribers, with plans for a gradual rollout to other subscription tiers. Q: What are the potential challenges for Wide Research? A: Challenges include the need for empirical evidence of its benefits over simpler methods, the complexity of multi-agent coordination, and navigating regulatory scrutiny due to its autonomous nature. Q: What pricing tiers are available for Manus? A: Manus offers different tiers, including Pro ($199/month), Plus ($39/month), and Basic ($19/month), with Wide Research initially available to Pro users.Source: Manus AI Launches ‘Wide Research,’ Pitting 100-Agent Swarms Against ‘Deep Research‘ from Google and OpenAI