Identity of Europe’s Leading AI Model Revealed: Quasar 438B Built on Zhipu GLM-5.2

Insight Beating AI Flash News: Shortly after Spanish AI firm Multiverse Computing billed Quasar 438B as “Europe’s strongest AI model”, the community uncovered its connection to Zhipu GLM-5.2. Artificial Analysis has now directly labeled the model as “Quasar 438B (max, based on GLM-5.2)”, explicitly noting it is built on GLM-5.2. Clues were also present in Multiverse’s old documents: when Quasar 438B launched its beta on August 5, CompactifAI’s update log stated its capabilities were “exactly identical” to GLM-5.2. The community is now questioning whether Quasar is merely a renamed, compressed or modified version of GLM-5.2. Multiverse has not yet disclosed its architecture, training methodology, or the exact relationship between the two models. Multiverse’s core business is model compression, yet its current FAQ still lists Quasar under “Original Models” rather than “Compressed Models”. GLM-5.2 has around 753 billion total parameters, with 40 billion activated during inference, scoring 53 points on Artificial Analysis’s max-tier intelligence index. Quasar, reduced to 438 billion parameters, only scores 43 points, but its output speed is approximately 174 tokens per second, significantly faster than GLM-5.2’s roughly 68 tokens per second.