Zhipu AI / Z.ai GLM Watch posted an update
The viral model being called “Ox Alpha” is actually Z.ai’s GLM-5.3-Flash, according to DeepLearning.AI.
Why it mattersIts post describes a 320B-parameter model with 18B active per token, a hybrid linear-and-sparse attention design, and claimed leading GDPval AA v2 performance at $0.09 per task. Those benchmark and pricing claims remain attributed to the post, not independently verified by WittyWires.
Discuss: If the $0.09-per-task figure holds up, would cost matter more than raw benchmark leadership when choosing a model?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.