Multiverse Computing is pushing its compressed AI models into the mainstream

With private equity firms operating at over 9.2% – the highest rate in years – VC firm Lux Capital recently advised companies relying on AI to get their computing commitments underwritten. As financial volatility erupts in the AI ​​supply chain, Lux warned, a handshake agreement is not enough.

But there is another option altogether, which is to stop relying on external computing infrastructure altogether. Small AI models that run directly on the user’s own device — no data center, no cloud provider, no third-party risk — are getting good enough to be considered. And Multiverse Computing raises its hand.

The Spanish startup has so far kept a lower profile than its peers, but as demand for AI efficiency grows, this is changing. After compressing models from major AI labs including OpenAI, Meta, DeepSeek and Mistral AI, it launched both an application that showcases the capabilities of its compressed models and an API portal – a gateway that allows developers to access and build on those models – making them widely available.

The CompactifAI app, which shares its name with Multiverse’s quantum-inspired compression technology, is an AI chat tool along the lines of ChatGPT or Mistral’s Le Chat. Ask a question, and the model will answer. The difference is that Multiverse is embedded Gilda, a model so small that it can work both online and offline, according to the company.

For end users, this is a taste of AI at the edge, with data that doesn’t leave their devices and doesn’t require a connection. But there is a caveat: their mobile devices must have enough RAM and storage space. If they don’t — and most older iPhones won’t — the app switches to cloud-based models with an API. The path between cloud processing is handled automatically by a Multiverse system named Ash Nazg, a name that will ring a bell for Tolkien fans as it refers to the One Ring text in “The Lord of the Rings.” But when an application moves to the cloud, it loses its primary privacy edge in the process.

These limitations mean that CompactifAI is not yet ready for mass customer adoption, although that may never be the goal. According to data from Sensor Tower, the app had less than 5,000 downloads last month.

The real target is businesses. Today, Multiverse is launching a self-service API portal that gives developers and businesses direct access to its compressed models — no AWS Marketplace required.

“CompactifAI API Portal 1773909494 it gives developers direct access to transparently compressed models and the control needed to move them into production,” CEO Enrique Lizaso said in a statement.

Real-time usage monitoring is one of the key features of the API, and that’s not a risk. Along with the potential benefits of edge deployment, lower computing costs are one of the main reasons why businesses are considering micro-models as an alternative to large-scale language models (LLMs).

It also helps that the smaller models are less restrictive than before. Earlier this week, Mistral updated its small model family with the launch of the Mistral Small 4, which it says is optimized for general chat, coding, agent tasks and reasoning. The French company also released Forge, a program that allows businesses to build custom models, including small models from which they can choose the tradeoffs their use cases can best tolerate.

The latest Multiverse results also suggest that the LLM gap is narrowing. Its latest compressed model, the HyperNova 60B 2602, is built on gpt-oss-120b – an OpenAI model whose base code is publicly available. The company says it now delivers faster responses at a lower cost than its original predecessor, a significant benefit in the flow of agent code, where AI automatically completes complex, multi-step tasks.

Making models small enough to work on mobile devices while still being useful is a big challenge. Apple Intelligence has avoided that problem by combining an on-device model with a cloud model. The Multiverse CompactifAI app can also submit requests to gpt-oss-120b via API, but its main goal is to show that local models like Gilda and their future replacements have benefits that go beyond cost savings.

For employees in critical locations, a model that can run locally and without connecting to the cloud offers more privacy and resilience. But the biggest value is in the business use cases this could open up – for example, embedding AI in drones, satellites, and other settings where connectivity can’t be taken for granted.

The company already serves more than 100 international clients including the Bank of Canada, Bosch and Iberdrola, but expanding its client base could help it unlock more funding. After raising a Series B of $215 million last year, it is now rumored to be raising a new funding round of 500 million euros at a value of more than 1.5 billion euros.