European artificial intelligence startup Multiverse Computing has announced the launch of two breakthrough ultra-compact AI models, SuperFly and ChickBrain, designed to operate offline while retaining strong performance across text generation and reasoning tasks.
The models are powered by the company’s proprietary CompactifAI compression technology, which applies quantum-inspired methods to drastically reduce model size and cost without sacrificing usability.
Tiny Models with Significant Power
Multiverse’s new releases are aimed at solving one of AI’s biggest challenges: how to run advanced models on devices with limited storage and processing power.
- SuperFly has just 94 million parameters, roughly the size of a fly’s brain, yet can generate text and perform reasoning at levels typically reserved for larger models.
- ChickBrain, compressed to 3.2 billion parameters from a standard 18 billion, performs well on key benchmarks such as MMLU-Pro, Math500, GSM8K, and GPQA Diamond.
The company refers to these innovations as part of its “Model Zoo”, a collection of highly efficient models that balance performance with accessibility.
CompactifAI: Quantum-Inspired Compression
At the core of these advancements lies CompactifAI, Multiverse’s proprietary compression method. The technology, informed by quantum algorithms and tensor networks, delivers several advantages:
- Reduces model size by up to 95%, enabling deployment on phones, PCs, and IoT devices.
- Cuts inference costs by up to 80%, making large-scale AI rollouts financially sustainable.
- Allows offline usage, addressing privacy concerns while removing dependence on cloud infrastructure.
These features could prove transformative for industries where data security, low latency, and affordability are crucial—particularly in healthcare, finance, telecommunications, and industrial automation.
Backed by Major Funding
The unveiling of SuperFly and ChickBrain follows Multiverse’s recent €189 million (US $217 million) Series B funding round, led by Bullhound Capital with participation from HP Tech Ventures, Toshiba, and other global investors. The capital is expected to accelerate the commercialization of CompactifAI and expand the company’s reach across new markets.
Founded in 2019 and headquartered in San Sebastián, Spain, Multiverse has established itself as one of Europe’s leading AI startups, particularly in applying quantum algorithms to enterprise use cases. Its shift toward compact model deployment reflects a growing demand for practical AI solutions that can operate outside large data centers.
Why This Matters
The launch of these compact models signals a shift in how AI could be deployed globally:
- Edge AI Enablement
Running advanced models locally on devices unlocks opportunities in low-connectivity environments and for sensitive applications requiring on-device processing. - Cost and Energy Efficiency
By lowering resource requirements, these models make AI more scalable and environmentally sustainable. - Expanded Use Cases
From offline customer support assistants to embedded AI in smart appliances, the possibilities expand beyond cloud-based applications. - European Leadership in AI Innovation
With this milestone, Multiverse positions itself as a strong competitor to US and Asian AI firms, showcasing Europe’s growing role in AI research and commercialization.
The Road Ahead
While these models do not rival the largest AI systems on raw accuracy, their strength lies in efficiency and accessibility. By focusing on smaller, deployable models, Multiverse is addressing a critical need in AI adoption: making the technology available to more organizations and devices without prohibitive costs.
As the AI race increasingly shifts toward practical deployment, Multiverse’s innovations highlight a growing trend—smaller may indeed be smarter.