Multiverse Computing has released "Quasar 438B," a inference model with over 400 billion parameters. The model supports English and Spanish.
Model ReleasesMultiverse ComputingQuasar 438B
Multiverse Computing Announces inference Model Quasar 438B, Recording Highest Score for a European Model
This article is a translation. Read the Japanese original
Quasar 438B achieved a score of 43 on the Artificial Analysis Intelligence Index v4.1.1. This is the highest result for a European model, surpassing the 30 recorded by Mistral Medium 3.5 and the 38 recorded by NVIDIA Nemotron 3 Ultra.
The model also features notable response speeds. The time taken to generate 500 tokens, including thinking time, was 15.3 seconds. While three of the compared models were faster, all of them had lower index scores than Quasar.
Specifically, it was shown that Gemini 3.7 Flash is both faster and has a higher index score than Quasar.
In the AA-LCR, which indicates long-context processing capability, the model recorded 75.0. This figure is comparable to state-of-the-art models such as Claude Opus 5. Additionally, it recorded a score of 69.3 on Terminal-Bench v2.1, which evaluates performance in terminal environments.
The model is available via the CompactifAI API. It is intended for use in software development agents, research systems, and workflow automation.
Source: Quasar 438B: Europe's Leading AI Model (Hacker News Frontpage, 2026-09-02)