Genspark Releases Manus-like Universal AI Agent: Genspark Super Agent

AIbase基地

Published inAI News · 8 min read · Apr 3, 2025

399

April 2, 2025, California – The competition in the artificial intelligence field is heating up. Following the launch of the highly anticipated general-purpose AI agent Manus by Chinese startup Butterfly Effect, Genspark, a US company founded by former Baidu executive Jing Kun, today announced its new product, "Genspark Super Agent," touted as a "fast, accurate, and controllable" general-purpose AI agent. This news quickly sparked heated discussions within the tech community, with many professionals drawing comparisons to Manus and viewing it as signaling a new round of competition in general-purpose AI agent technology.

Genspark Super Agent: A Sneak Peek at Key Features

According to recently released information, Genspark Super Agent not only rivals Manus in performance but also introduces unique features, aiming to stand out in the competition. Tech bloggers in their reviews point out that Genspark's general-purpose agent has a built-in AI module capable of automatically making phone calls, a particularly user-friendly feature for international users. For example, it can help users automatically book hotels or restaurants, significantly enhancing its practicality. Industry insiders have also revealed that this agent is powered by a robust "Mixture-of-Agents" system, integrating 8 LLMs (Large Language Models) of varying sizes, over 80 toolkits, and a massive amount of high-quality datasets, enabling it to handle various complex tasks.

twitter_Genspark(@genspark_ai)_20250402-154911_1907460308362523114_photo.png

twitter_AINOSUKE @AIクリエイター(@ainosuke2021x)_20250402-161052_1907465767358222354_photo (1).jpg

Furthermore, the official demo video showcases the Genspark Super Agent's performance in call booking scenarios. In the video, the AI agent not only understands user needs but also smoothly communicates with service providers over the phone, ultimately completing the booking process. The unveiling of this feature has amazed many with its level of automation and is seen as a significant step beyond traditional AI assistants.

Genspark vs. Manus: Competition or Complementarity?

Since the launch of Manus on March 6, 2025, its title as the "world's first fully autonomous AI agent" has garnered widespread attention. Manus is known for its multi-agent architecture and ability to independently perform complex tasks such as screening resumes, analyzing stock trends, and generating interactive websites. The release of Genspark Super Agent seems to be a deliberate attempt to compete in this space. Industry experts have shared detailed information about Genspark, compiling it into systematic analyses to help more people understand its capabilities.

Although both are categorized as general-purpose AI agents, based on currently available information, Genspark Super Agent appears to be optimized for interactivity and practicality. The built-in phone functionality is considered one of its key differentiators from Manus. A technology reviewer noted: "Today, the general-purpose agent field has a new player. Genspark released their Manus-like general-purpose agent. The difference is that they have built-in AI phone capabilities, which are very convenient for international users." This view is shared by industry experts, with many researchers believing this feature fills a gap in Manus's real-world application scenarios.

Industry Response and Future Outlook

The launch of Genspark Super Agent comes at a crucial time for the rapid development of AI agent technology. Earlier this year, Manus surpassed OpenAI's Deep Research in the GAIA benchmark test, sparking heated discussions about the rise of Chinese AI technology. Genspark's entry into the market undoubtedly injects new vitality into this field. Some industry analysts predict that this new agent may further challenge existing technological standards in terms of speed and reliability.

However, some voices have also pointed out that the specific performance of Genspark Super Agent remains to be verified. Researchers in technical forums mention that the details of its "Mixture-of-Agents" system have not been fully disclosed, and whether its real-world performance can live up to its advertised "fast, accurate, and controllable" claims remains to be seen. Meanwhile, Manus's development team has also indicated plans to open-source parts of its model within the year, which may put even greater competitive pressure on Genspark.

Conclusion

With the appearance of Genspark Super Agent, the general-purpose AI agent field is becoming increasingly crowded. From Manus to Genspark, these technological breakthroughs not only showcase AI's transformation from a simple conversational tool to an agent capable of independently executing tasks, but also foreshadow a profound change in how humans and machines will collaborate in the future. As industry discussions reflect, professionals and users alike are closely watching the performance of these emerging agents, anticipating that they will bring more surprises in real-world scenarios. The competition between Genspark Super Agent and Manus over the next few months will undoubtedly be a major highlight in the AI field.

Learn more about Genspark: https://www.aibase.com/zh/tool/31186

NVIDIA Unveils Multimodal LLM Describe Anything: Generating Detailed Descriptions of Specific Regions

The NVIDIA AI team has released a revolutionary multimodal large language model—Describe Anything 3B (DAM-3B)—designed for detailed, region-specific descriptions of images and videos. This model, with its innovative technology and superior performance, has generated significant discussion in the multimodal learning field, marking another milestone in AI development. Below, AIBase outlines the model's core highlights and industry impact. A breakthrough in region-specific descriptions, DAM-3B stands out for its unique ability to...

Hong Kong and UK Research Teams Propose Innovative Image Tokenization Method; Hierarchical Structure Improves Reconstruction Quality

Researchers from Hong Kong and the UK have recently proposed a novel image tokenization method designed to convert images into digital representations (tokens) in a more compact and accurate manner. Unlike traditional methods that distribute information uniformly across all tokens, this method employs a hierarchical structure, capturing visual information layer by layer, thereby improving the quality and efficiency of image reconstruction. Traditional image tokenization techniques typically divide each part of an image equally into multiple tokens, whereas the new method adopts a hierarchical structure. Initial tokens encode rough shapes and structural elements,

OpenAI Releases gpt-image-1 API: 4o Image Generation Capabilities Now Open

OpenAI has officially launched the gpt-image-1 API, marking the opening of its highly anticipated 4o image generation capabilities to developers. According to AIbase, this API is lauded by the community as the world's strongest 'image generation' tool due to its high-fidelity image generation, diverse visual styles, and powerful integration of world knowledge. The release announcement has generated significant excitement among AI developers and the creative community, with relevant documentation now publicly available via the OpenAI website and Playground platform. Core features: High-fidelity and diverse style generation

Microsoft 365 Copilot Adds Researcher and Analyst Bots to Boost Enterprise Intelligence

Microsoft recently updated its 365 Copilot with new features, including AI-powered enterprise search, a reasoning agent, and a new Copilot store. Some users have access to these features now, while others may need to wait until May. The AI-powered enterprise search is a particular highlight. Microsoft states that Copilot search combines search with the power of AI to deliver more relevant and context-aware search results. This is from the Vice President of Microsoft's design and research department.

AWS Releases SWE-PolyBench: A New Open-Source Benchmark for Evaluating AI Programming Assistants

AWS AI Labs recently introduced SWE-PolyBench, a multilingual open-source benchmark designed to provide a more comprehensive framework for evaluating AI programming assistants. With advancements in large language models (LLMs), AI programming assistants capable of generating, modifying, and understanding software code have shown significant progress. However, current evaluation methods remain limited, with many benchmarks focusing solely on single languages like Python, failing to offer a complete picture.

Revolutionizing Tradition! Hailing Group and Huawei Launch the Cement Industry's First AI Large Model to Empower Intelligent Upgrades

A recent technological launch event has garnered significant attention within the cement and building materials industry. Hailing Group, a leading domestic cement manufacturer, partnered with tech giant Huawei to unveil the industry's first AI large model. This is not merely a simple technological application, but a profound revolution of traditional production methods. This AI large model is reportedly built upon the powerful Huawei Cloud PanGu series of large models. It is not theoretical; it is deeply integrated into real-world cement production scenarios, comprehensively covering aspects such as quality control.

OpenAI Predicts $125 Billion Revenue by 2029, 3 Billion Monthly Active Users by 2030

OpenAI recently released a prediction forecasting $125 billion in total revenue by 2029. AI agent and channel revenue will be key drivers. AI agent revenue is projected to reach nearly $29 billion, representing almost a quarter of total revenue, while channel revenue is expected to reach $25 billion. Image note: Image generated by AI, image licensing service Midjourney. Following the success of ChatGPT, OpenAI's...