Forget chasing Manus invites; the focus has shifted to the open-source community! Recently, the CAMEL-AI team proudly announced the launch of their groundbreaking OWL (Optimized Workforce Learning) project. This innovative multi-agent collaboration framework, built on the CAMEL-AI framework, is making waves in the automated task processing field with its exceptional performance and open nature.
Topping the GAIA Leaderboard
OWL's capabilities speak for themselves. Officially, OWL achieved a remarkable average score of 58.18 on the renowned GAIA benchmark test, securing the top spot among open-source frameworks.
This score even surpasses Huggingface's Open Deep Research. This begs the question: what about the legendary "cloud-based super worker," Manus? The CAMEL-AI team's response is playfully geeky: "Is that all? We'll replicate it in 0 days!"
Open-Source Sharing
Unlike Manus, still in mysterious closed beta with invites fetching exorbitant prices, OWL's biggest highlight is its complete open-source nature.
Developers can directly clone the code on GitHub, experience this powerful framework firsthand, and even contribute to OWL's development, collaboratively shaping a more powerful, all-purpose open-source agent.
The project address is conveniently provided: https://github.com/camel-ai/owl.
Feature Highlights: Intelligent Performance Beyond Expectations
So, what makes this open-source OWL dare to challenge Manus?
- Flexible and Efficient Multi-Agent Collaboration: OWL's core lies in dynamic agent interaction for more natural, efficient, and robust task automation. It's not just about multiple AIs working together; it emphasizes collaboration models and communication protocols.
- Powerful Task Automation Capabilities: OWL aims to push the boundaries of task automation. Its goal is to revolutionize how AI agents collaboratively solve real-world problems.
- Easy Installation and Configuration: The CAMEL-AI team provides clear installation instructions. Whether using Conda or venv, developers can quickly set up OWL's runtime environment and configure API keys as guided.
- GAIA Benchmark Test Reproduction: OWL also provides scripts for reproducing the GAIA benchmark test results, allowing developers to verify its performance.
- Ambitious Future Plans: The CAMEL-AI team has outlined OWL's future development, including writing technical blogs, enhancing the tool ecosystem, and developing more complex agent interaction models.
By studying Manus's technological approach, OWL is actively replicating and surpassing its functionality. The CAMEL-AI team has launched the Workflow Replication Project and the Ubuntu Toolkit Replication Project. This means that what Manus can do, such as launching Ubuntu containers, connecting data sources, automatically generating task lists, executing terminal commands, parsing files, and manipulating browsers, OWL is actively implementing, and will be even more flexible.
Even more exciting is the integration of CAMEL-AI's previously developed cross-platform agent CRAB technology into OWL. This means that the future OWL will not only be able to control cloud containers but also potentially directly control applications on phones and computers, achieving true cross-platform, multi-device, all-scenario remote operation, becoming a true open-source Manus Pro Max.
