Multiverse Computing launches Quasar 438B for enterprise AI

Multiverse Computing has launched Quasar 438B, a flagship reasoning model that strengthens Europe’s position in the global AI race. The Spanish company’s 438-billion-parameter model scored 43 on the Artificial Analysis Intelligence Index v4.1.1, ahead of Mistral Medium 3.5 on 30, NVIDIA Nemotron 3 Ultra on 38 and Inkling on 42. Claude Opus 5 leads the current comparison with 63.

Quasar is built for enterprise-scale agents and coding rather than consumer chatbot use. It is designed for multi-step work involving planning, tool use, code execution and large volumes of information, operates in English and Spanish, and is available through Multiverse Computing’s CompactifAI API. Multiverse Computing says Quasar produces 500 output tokens in 15.3 seconds, including reasoning time, making speed a central part of its enterprise pitch.

The model also scored 75.0 on Artificial Analysis Long Context Reasoning, matching Grok 4.6, while its Terminal-Bench v2.1 score of 69.3 points to practical coding and terminal-based agent performance. Quasar does not close Europe’s gap with the leading US-developed systems, but it shows European developers competing on measurable performance, efficiency and commercially deployable AI infrastructure.