StartLux's local large model passes key test

The local large model developed by Shanghai-based company StartLux has passed the test of the China Academy of Information and Communications Technology, marking a significant breakthrough.

According to the academy's report, in the model context protocol test, StartLux's local large model ranked second in overall performance and first in multiple intelligent agent tasks. This represents a qualitative breakthrough in the ability of local models to complete tasks, and the competitiveness of Chinese artificial intelligence continues to strengthen.

The test covered six specialized tasks and comprehensive evaluations, namely location navigation, web search, browser automation, financial analysis, code repository management, and 3D design. There are a total of 7 inspection items, with a focus on examining the model's tool calling, multi-step planning, and task completion abilities in real work scenarios.

Chuan-Wei Kuo, chief technology officer of StartLux, said that the model can run on consumer-grade personal computers, which means that ordinary consumers can use AI with strong capabilities without expensive professional equipment. This is of great significance for promoting the popularization of AI in various industries such as education, healthcare, scientific research, office, and people's daily lives.

StartLux plans to launch the first generation of local intelligent solutions within this year, which users can directly download and use without complex deployment and debugging.