Voices
Zhipu's anonymous 'Ox Alpha' is confirmed as GLM-5.3-Flash: 320B total, 18B active, with Databricks claiming 10% better quality at 1/10 the cost of GLM-5.2
The model that had been running unnamed on OpenRouter is Zhipu's GLM-5.3-Flash, per the August 28 AINews roundup, and Unsloth says a 3-bit GGUF runs on 128GB of RAM. Databricks' claim of 10% higher quality than GLM-5.2 at a tenth of the cost is vendor-supplied and unverified independently. Treating it as confirmed depends on a single aggregator, so hold the numbers loosely until Zhipu publishes a card.
↳ Follow the thread