Ant Ling launches Ling-3.2-flash, a huge multimodal model with 1M token context

Ant Group's Ling team released Ling-3.2-flash, a native multimodal model that can take in images, text, and video. It has 560 billion total parameters, though only 25 billion are active at once, and supports a context window of up to 1 million tokens. The company built it for tasks like detailed visual understanding, long-running multimodal agents, 3D understanding and interaction, and medical multimodal analysis. Developers can try it through Ant Ling's chat interface or connect to it using APIs compatible with both OpenAI and Anthropic formats, making it easy to swap into existing tools. For the first two weeks after launch, the model is free to use, giving developers a window to test it before any charges apply. The release adds to a growing list of large multimodal models competing on context length and parameter efficiency, with Ling-3.2-flash notable for combining a very large total parameter count with a much smaller active count, a design meant to keep inference costs down while still handling complex, long-horizon tasks across multiple data types.

Why it matters

A free, large-context multimodal model lowers the barrier for building agents that reason over long video, image, and text inputs together.

Source: Ant Ling (Ant Group / inclusionAI)