Alibabas Qwen3 AI powers DeepSWE to top global ranking
Alibabas Qwen3-32B model drives DeepSWE to achieve 59 percent accuracy in the latest SWEBench-Verified test, outperforming DeepSeek-V3.
Alibaba Group Holding’s open-source Qwen artificial intelligence (AI) model has enabled the agentic framework DeepSWE to outperform rival systems in the growing AI agent segment.
Key Highlights:
- DeepSWE, jointly developed by Agentica and Together AI, was trained on Alibaba’s Qwen3-32B large language model (LLM).
- It achieved 59% accuracy in the SWEBench-Verified test, surpassing competitors like DeepSeek-V3-0324.
- The framework marks Alibaba’s expanding leadership in the open-source AI community.
What Are Agentic Frameworks?
Agentic frameworks are software platforms that provide tools to build, deploy, and manage AI agents. These agents, such as Butterfly Effect’s Manus, autonomously perform tasks by creating plans and subtasks to achieve goals.
Open-Source Advantage
Alibaba’s open-source approach allows public access to its AI models, enabling third-party developers to modify, fix, or scale capabilities. This strategy has propelled Qwen3 to the top of global rankings, solidifying Alibaba’s role in advancing AI innovation.
Alibaba owns the South China Morning Post.
Related News
SF AI Meetup Explores Next Gen Autonomous Agents and ML
SF AI/ML Meetup on Engineering Next Generation AI Systems with autonomous agents and ML architectures featuring industry leaders.
Salesforce shares key lessons from Agentforce AI rollout
Salesforce EVP Joe Inzerillo reflects on major insights and pitfalls from the company's first year using its Agentforce AI platform
About the Author
Alex Thompson
AI Technology Editor
Senior technology editor specializing in AI and machine learning content creation for 8 years. Former technical editor at AI Magazine, now provides technical documentation and content strategy services for multiple AI companies. Excels at transforming complex AI technical concepts into accessible content.