The Artificial Analysis Intelligence Index score for a new large language model has risen to 44, up from a previous score of 41. AntLingAGI introduced the Ling-3.1-flash-1019 model, which also showed gains in agentic tasks with its AA-Briefcase score rising from approximately 1400 to 1524 Elo and AutomationBench-AA increasing from 62% to 68%. The Chinese lab will release API access and open weights for the text-only model shortly.
The model employs a Mixture-of-Experts architecture with 560B parameters, 25B of which are active. The newcomer rivals the performance of DeepSeek-v4.1 Flash-1710 and GLM-5.3 despite having a smaller footprint. AntLingAGI is an emerging player in the Chinese AI sector and currently maintains a following of 3,000 users on Hugging Face.
