AntLingAGI Debuts No. 2 Open Weights Model for Mobile Apps

The new AI model from AntLingAGI achieved a 1,207 Elo rating and a 17th place overall ranking on the Design Arena's leaderboard for app development. Ling-3.1-flash ranks as the No. 2 open-weights model in that category, demonstrating specific proficiency in creating games and mobile productivity trackers with performance comparable to Opus 4.6. It also recorded a 1,673 Elo on GDPVal-AA v2.1, 75.16 on FrontierSWE, and 65.35 on HealthBench Professional.

AntLingAGI plans to open-source the model, which features approximately 560B total parameters and 25B active parameters per token with a 1M-token context window. While benchmark scores on key tests are close to GPT-5.6 sol and Opus 5, some analysts have noted that the model has not yet been compared to more recent versions such as Opus 5.5 or GPT-6.1 sol.