Ant releases Ling 3.2 Flash: in-house tests beat Claude Opus 5, free for the first two weeks
Ant’s Ling team has released Ling-3.2-flash, a natively multimodal model with 560B total parameters, 25B active, and a 1M-token context window, supporting image, text and video input. In Ant’s own evaluations it outperformed Claude Opus 5 (max) on OmniDocBench, ZEROBench and Vision2Web, and it is free to try for the first two weeks. It matters because a major Chinese lab is claiming flagship-level multimodal performance with OpenAI- and Anthropic-compatible APIs, intensifying competition at the top of the model race — though the benchmarks are self-reported.
SOURCES
- IT之家2026-11-06 00:00