I heard about the tools of “deep research” of artificial intelligence … Now Manus launches “wide research” that revolves more than 100 agents to achieve the web for you

Chinese start starting Manusany Newspaper headlines are published earlier this year For its approach to the multi -agent coincidence platform for consumers and “professionals” (professionals who want to operate work operations), it is due to an interesting new use of its technology.

While many major artificial intelligence providers are like Openaiand GoogleAnd xi The “deep research” or “deep researcher” has launched artificial intelligence factors who make minutes or hours of extensive and in -depth research on the Internet and writing comprehensive reports well -used on behalf of users, and Manus takes a different approach.

the The company has just announced “broad research”, A new experimental feature that enables users to carry out large tasks and size by taking advantage of the strength of parallel artificial intelligence agents-even more than 100 at one time, all focus on completing one task (or a series of sub-tasks that hand over the aforementioned goal).

It was previously reported that it uses the HWEN models to run its platform.

Parallel treatment for research, summary and creative output

in The video was posted on the official X accountThe co -founder of Manus and the chief scientist Yichao ‘Peak’ Ji shows an experimental offer for extensive research to compare 100 sports shoes.

To complete the task, the extensive research revolves around Manus almost immediately 100 simultaneous branches – each of which was assigned to analyze the design of one shoe, its prices and its availability.

The result is a sorting matrix that is delivered in each of the spreadsheets and the web page within minutes.

The company suggests that extensive research is not limited to data analysis. It can also be used for creative tasks such as exploring design.

In one of the scenario, Manus agents were created simultaneously, sticker designs across 50 distinct visual patterns, and refined assets in the Zip fileable file.

According to Manus, this flexibility stems from the system at the system level for parallel treatment and communication, an agent.

In the video, Peak explains that the wide search is the first application for improved virtual simulation and architecture capable of limiting the 100 -time account capacity that exceeds the initial offers.

The feature is designed to activate automatically during tasks that require a large -scale analysis, with no manual switch or required configurations.

Availability and pricing

Wide research is available from users in the Manus Pro plan and will be gradually available for those in basic and basic plans. As of now, the Manus subscription pricing has been organized as follows a month.

• free – It includes $ 0 per month 300 daily updates, access to a chat mode, one simultaneous task, and one scheduled task.
• essential – A monthly dollars will add 1900 credit hours (+1,900 bonus during the limited offer), 2 simultaneous and renewable missions, access to advanced models in the agent position, images/video/video generation, and exclusive data sources.
• plus – It exceeds 39 dollars per month to 3 simultaneous and 3 scheduled tasks, and 3,900 credit hours (+3,900 bonus), and includes all the basic features.
• Professional -199 dollars/month provides 10 synchronous tasks, 10 scheduled tasks, 19,900 credit hours (+19,900 bonus), early access to trial versions features, Manus shirt, and full features of features including advanced agent tools and content generation.

There is also a 17 % discount on these prices for users who want to pay off annually.

The launch depends on the infrastructure that was presented with Manus earlier this year, which the company describes as not just the agent of Amnesty International, but a personal cloud computing platform.

Each Manus session works on a dedicated virtual device, allowing users to access the organized cloud account through the natural language-a preparation that the company considers a key to enabling AI’s workflow for general purposes.

Through extensive research, Manus users can delegate research or creative exploration across dozens or even hundreds of subcontractors.

Unlike traditional multi-agent systems with pre-determined roles (such as the manager, programmer or designer), each sub-group within wide research is a fully capable Manus, not a specialist for a specific role-works independently and able to take over any general task.

The company says that this architectural decision opens the door to flexible and developed tasks that are not restricted to solid templates.

What are the benefits of deep research?

The implicit meaning seems to be that the operation of all these factors in parallel faster and will lead to a better and more varied group of work products outside research reports, unlike the individual “deep research” factors that were shown by other artificial intelligence providers or are offered.

But while Manus enhances extensive research as a penetration in parallel with the agent, the company does not provide direct evidence that spawning dozens or hundreds of subcontractors are more effective than having a single -capacity treatment tasks successively.

The version does not include performance criteria, comparisons, or technical interpretations to justify the bodies of this approach-such as increasing the use of resources, complexity of coordination, or potential shortcomings. It also lacks details on how the internalists cooperate, how to combine results, or whether the system provides measurable advantages in speed, accuracy or cost.

As a result, while the advantage shows the architectural ambition, its practical benefits remain on simpler methods that are not proven based on the information provided.

The sub -factors have a generally mixed record in general, so far …

While Manus’ implementation of extensive research is placed as advanced in the systems of general artificial intelligence agent, the broader ecosystem has witnessed mixed results with similar methods.

For example, on Reddit, self -described users of Claude’s symbol Fears about its slow, consuming large quantities of symbols, and providing a limited vision for implementation.

Common pain points include the lack of coordination protocols between factors, difficulties in correcting errors, and irregular performance during high load periods.

These challenges are not necessarily reflected in Manus’s implementation, but they highlight the complexity of the development of strong multi -agent frameworks.

Manus admits that extensive research is still experimental and may come with some restrictions with continued development.

We look forward

With extensive research, Manus deepens its commitment to redefining how users interact with artificial intelligence agents on a large scale.

While other platforms wrestle with the technical challenges of the coordination and reliability of Subagent, the manus approach may be a test of whether the agents of the agent-instead of the stereotypes that were narrow-can meet the vision of AI’s cooperation and multi-thread.

The company alludes to broader ambitions, indicating that the infrastructure behind the extensive research lay the basis for future shows. Users and industry scenes will pay close attention to whether this new wave of the agent’s structure can rise to the level of its capabilities – or whether the challenges seen elsewhere in the area of artificial intelligence will eventually take.