On December 19, 2024, at a press conference, ZhiYuan Research Institute and Tencent announced the launch of LongBench v2, a benchmark designed to evaluate the deep understanding and reasoning capabilities of large language models (LLMs) in real-world long text multi-task scenarios. The platform aims to advance long text models in understanding and reasoning, addressing the current challenges faced by large language models in practical applications.

Long context understanding is a key challenge in the field of natural language processing, especially when large language models (LLMs) handle text that exceeds their context window size. To address this issue, researchers developed the LooGLE benchmark test, aimed at assessing LLMs' capability to understand extremely long documents (averaging 19.3k words, comprising 776 articles across multiple domains). LooGLE includes 7 tasks that cover both short and long dependencies, evaluating model understanding across texts of varying lengths. The test data is sourced from open-access content after 2022.

Google has launched a new AI system called Gemini, which excels in benchmark testing, outperforming existing models. The Gemini system, developed by Google DeepMind, supports multiple modalities including text, images, audio, video, and code. There are three versions of the Gemini system optimized for different use cases. Compared to competitors, the Gemini system has made progress in narrowing performance gaps. The multifunctional features of the Gemini system bring commercial potential in areas such as cloud services.

Amazon AWS has launched a Human Benchmark Testing team to enhance AI model evaluation. Amazon aims to help users better assess AI models and encourages more people to participate in this process. AWS provides model evaluation on Bedrock to assess models in its repository. Model evaluation consists of both automated and human assessments, which can evaluate model performance based on different metrics. AWS also offers a human evaluation team to collaborate with users and detect metrics that automated systems cannot.

The first systematic benchmark test for AI agents has been released, showcasing comprehensive evaluation results for 25 different language models: GPT-4 stands out uniquely. Top commercial language models perform excellently in complex environments, showing significant advantages over open-source models. The research team suggests further improving the learning capabilities of open-source models.