LISABench has kicked off 2026 with its first major AI security benchmark test, focusing on Web3 smart contract vulnerability detection. The Q1 2026 evaluation features seven leading AI models, including KIMI K2, DeepSeek V3.2, and GPT-5.2, competing to identify vulnerabilities in real-world scenarios. In addition to the technical evaluation, LISABench has introduced a community voting prediction reward program, allowing participants to predict the top-performing model. The benchmark's codebase is now open-sourced on GitHub, enabling developers to review and replicate the tests. Voting for the Q1 winner is currently open.