Multiverse Computing’s Quasar 438B Tops European AI Models on Intelligence Index
Event summary
- Multiverse Computing launched Quasar 438B, a 438-billion-parameter model scoring 43 on the Artificial Analysis Intelligence Index v4.1.1, the highest for a European model.
- Quasar outperforms Mistral Medium 3.5 (score 30) and NVIDIA Nemotron 3 Ultra (score 38) while producing 500 tokens in 15.3 seconds.
- The model leads in long-context reasoning (75.0 on AA-LCR) and coding (69.3 on Terminal-Bench v2.1), surpassing key competitors in enterprise-relevant tasks.
- Quasar is available via the CompactifAI API, supporting English and Spanish for cross-border enterprise deployment.
The big picture
Multiverse Computing’s Quasar 438B marks a strategic milestone for European AI, proving that continent can compete with US and Chinese models in both capability and speed. The launch underscores growing demand for sovereign, high-performance models in enterprise settings, particularly where latency and reasoning accuracy are critical. With over 100 global customers, including Iberdrola and Bosch, Multiverse is positioning itself as a key player in the compressed AI model space.
What we're watching
- European Sovereignty
- Whether Quasar 438B can sustain its lead as a benchmark for European AI, challenging US and Chinese dominance in enterprise reasoning models.
- Performance Scaling
- The pace at which Multiverse Computing improves Quasar’s coding and agentic capabilities, particularly against frontier models like Claude Opus 5.
- Enterprise Adoption
- How quickly European firms deploy Quasar for document-heavy research, software engineering, and operational automation, given its speed and reasoning balance.