Ant Ling releases Ling-3.2-flash multimodal MoE model

Ant Ling has released Ling-3.2-flash, a 560-billion parameter mixture-of-experts model with 25 billion active parameters that natively processes text, images, and video. Available via API at $0.30 per million input tokens and $0.90 per million output tokens with a two-week free tier, the model aims to reduce the costs of multimodal agent loops. According to vendor benchmarks, it surpasses Claude Opus 5 on OmniDocBench, Vision2Web, and ZEROBench Sub. Open article →