Europe's 'Strongest AI Model' Quasar 438B Exposed as Based on Zhipu GLM-5.2
Spanish AI company Multiverse Computing recently promoted Quasar 438B as "Europe's strongest AI model," but the community quickly discovered its connection to Zhipu's GLM-5.2. Artificial Analysis' page now directly labels the model as "Quasar 438B (max, based on GLM-5.2)," explicitly stating its foundation on GLM-5.2.
Multiverse's own old documents also provide clues. When Quasar 438B launched in beta on August 5, CompactifAI's changelog stated that the model's capabilities were "identical" to GLM-5.2. The community has questioned whether Quasar is simply a compressed or modified version of GLM-5.2 repackaged under a new name.
Multiverse's core business is model compression, yet its FAQ still lists Quasar under "Original Models" rather than "Compressed Models." The company has not disclosed the architecture, training methods, or explained the relationship between the two models.
Performance Comparison
Data comparisons show that GLM-5.2 has about 753B total parameters with 40B activated during inference, scoring 53 on Artificial Analysis' intelligence index max tier. Quasar, compressed to 438B, scores 43 but boasts an output speed of about 174 tokens per second, significantly faster than GLM-5.2's approximately 68 tokens per second.
Multiverse has not yet officially responded to the community's questions, leaving the controversy over Europe's top AI model identity unresolved.