Spanish deep-tech firm Multiverse Computing launched Quasar 438B on September 2, 2026, its first large model and now the highest-scoring European entry on the Artificial Analysis Intelligence Index. Quasar is a reasoning model built for coding and agent workflows, and it is available today through the company's CompactifAI API. It scores 43 on the index, ahead of every other model developed in Europe.
Try It: Wire Quasar Into a Coding Agent
Quasar is API-only, so the fastest path is to sign up for the CompactifAI dashboard and point an existing agent or copilot at the new endpoint. The model returns 500 output tokens in 15.3 seconds, with only three models in the Artificial Analysis comparison running faster, so it fits latency-sensitive loops like inline completion, test generation, or a terminal agent. If you already run a router, add Quasar as a fallback tier for jobs that need European data residency, then benchmark it against your current coding model on your own repo before moving production traffic.
Why It Matters for Creators
Builders who need a sovereign, EU-based model have had thin options. Quasar shifts that: per the launch announcement, it scores 43 against Mistral Medium 3.5 at 30 and NVIDIA Nemotron 3 Ultra at 38, and posts 69.3 on Terminal-Bench v2.1 for agentic coding. Native English and Spanish support makes it practical for bilingual products, and it slots next to the wave of efficient coding models we broke down in our GLM-5.3-Flash vs Qwen vs DeepSeek V4 comparison.
Key Details
Model: Quasar 438B, a reasoning model for enterprise-scale agents and coding, and Multiverse Computing's first large model.
Score: 43 on the Artificial Analysis Intelligence Index v4.1.1, the highest of any European model, with 75.0 on long-context reasoning (AA-LCR).
Access: Generally available now through the CompactifAI API. Multiverse is known for quantum-inspired model compression, which is how it targets the speed and cost profile detailed by The AI Insider.
What to Do Next
Sign up for the CompactifAI API and run Quasar against a real coding task you care about, then compare tokens-per-second and output quality with your incumbent. If cost is your constraint, this is the same efficiency story driving the broader price war we tracked in the frontier API price drop.
