[Ant Group Releases Ling 3.0 Flash Model, Artificial Analysis Score Rises to 38] According to Deep Tide TechFlow, on August 8, Ant Group released the open-source inference model Ling 3.0 Flash, which adopts a hybrid inference architecture with 124B total parameters and 5B active parameters. The model achieved a score of 38 on the Artificial Analysis intelligence index, an improvement of 24 points compared to the previous Ling 2.6 Flash, matching MiMo-V2.5 and Qwen 3.6 27B with fewer active parameters. The model is configured with a 262K context window, with API pricing set at $0.075 per million input tokens and $0.22 per million output tokens, and is open-sourced under the MIT license.