Multiverse Computing Launches Quasar 438B, Europe's Highest-Ranking AI Model

European AI company Multiverse Computing has launched Quasar 438B, a large reasoning model that the company says is the most intelligent AI model built in Europe. According to the company's announcement, Quasar 438B scores 43 on the Artificial Analysis Intelligence Index, the highest result of any European model measured on that benchmark.

The launch marks a significant step for the Spanish firm, which built its reputation on AI model compression rather than training frontier-scale systems. Quasar 438B is the first large model Multiverse Computing has released, and it arrives as European governments and companies push for AI capability that does not depend entirely on American and Chinese labs. For readers following our AI news coverage, the release also signals that the race to close the gap with frontier models is no longer limited to the usual US contenders.

What Quasar 438B Actually Scores

The Artificial Analysis Intelligence Index (version 4.1.1) combines nine evaluations, including GDPval-AA v2, tau3-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience and AA-LCR. Quasar 438B's composite score of 43 puts it ahead of Mistral Medium 3.5 at 30, NVIDIA's Nemotron 3 Ultra at 38, and Inkling at 42, per the company's announcement.

The wider field still belongs to the American frontier: Claude Opus 5 leads the index at 63. Multiverse Computing does not claim otherwise. Instead, the company frames Quasar 438B as the strongest European option, outperforming comparable European models on seven of the eight selected Artificial Analysis evaluations it highlighted.

Speed Is the Second Argument

Raw benchmark scores are only half of Multiverse Computing's pitch. The company says Quasar 438B returns 500 tokens, thinking time included, in 15.3 seconds. Among the models in its comparison, only Nemotron 3.5 Lightning (9.4 seconds), Gemini 3.5 Flash-Lite (10.8 seconds) and Gemini 3.7 Flash (11.5 seconds) are faster, and only Gemini 3.7 Flash is both faster and more capable on the index.

The comparison against Mistral Medium 3.5 runs one way on both axes: Quasar scores higher (43 against 30) and answers faster (15.3 seconds against 18.8 seconds). Inkling, which nearly matches Quasar's intelligence at 42, needs 48.3 seconds for the same 500-token response, more than three times as long.

Strong Long-Context Reasoning

One result stands out from the announcement: on AA-LCR, the evaluation that tests the ability to extract, connect and reason over information spread across long documents, Quasar 438B scores 75.0. That is level with Grok 4.6 (high), within a point of Claude Opus 5 at 75.7, and ahead of Qwen3.8 2.4T A95B at 75.3, according to the company's figures.

Long-context handling matters because enterprise research, document analysis and agentic workflows are built on it. If a model cannot hold and connect information across a long document, it cannot power the multi-step tools companies actually want to deploy. This is where Quasar comes closest to the frontier group.

Agentic Coding Still Has Headroom

The weakest result in the announcement is Terminal-Bench v2.1, which puts coding agents to work in real terminal environments. Quasar 438B scores 69.3, leading Mistral Medium 3.5 by 18.7 points and Nemotron 3 Ultra by 15.4, but trailing the frontier group led by Claude Opus 5 at 89.1. Multiverse Computing acknowledges this is the evaluation with the most headroom and says the next round of work is aimed at it.

Built for Enterprise Agents

Quasar 438B is positioned as a model for organizations running software development agents, technical copilots, research systems and workflow automation. It sits in the 400-billion-parameter-plus class, handles English and Spanish, and is designed for multi-step tasks that need planning, tool use, code execution and large context without the latency that class normally carries.

Access runs through the company's CompactifAI API, so teams can test the model without standing up their own infrastructure. Multiverse Computing says more updates for Quasar are coming.

What It Means for the European AI Race

The release lands at a peculiar moment for European AI. Mistral, long the continent's standard-bearer, has faced questions about its direction, while US labs have consolidated their lead on composite benchmarks. A European model that legitimately tops the European field on an independent-ish index gives enterprises on the continent a credible regional option, particularly for bilingual English-Spanish deployments.

Whether Quasar 438B holds that position is another matter. The company itself notes that only a fraction of the index leaders answer faster, and the gap to Claude Opus 5 remains twenty points wide. But the compression-first company has now shown it can compete in the heavyweight class, and European enterprises finally have a home-market model worth benchmarking against their US defaults.

Stay Ahead of AI

The AI landscape shifts by the hour, and benchmark leads that looked unassailable in January are gone by summer. Read more AI news on AI Buzz Wire to track every model release, benchmark result and policy shift as it happens.

Follow the latest AI developments here.