Background and Market Context

Amid rising financial instability in the AI supply chain, companies are seeking ways to reduce reliance on external compute infrastructure. Multiverse Computing, a Spanish startup, has pursued a strategy of compressing large language models from major AI labs such as OpenAI, Meta, DeepSeek and Mistral AI. By making these models smaller, the firm aims to enable on‑device execution that offers privacy, resilience and cost savings.

CompactifAI App Launch

The company’s new CompactifAI app showcases its compression technology, dubbed CompactifAI, and features a model named Gilda. Gilda is small enough to run locally and offline on devices that have sufficient RAM and storage. When a device cannot meet the hardware requirements, the app automatically switches to cloud‑based models via an API, a process managed by a system called Ash Nazg. This hybrid approach means the app can provide a privacy‑first experience while still offering functionality on older hardware.

Self‑Serve API Portal

In addition to the consumer‑focused app, Multiverse has opened a self‑serve API portal that gives developers and enterprises direct access to its compressed models. The portal eliminates the need for third‑party marketplaces such as AWS Marketplace and includes real‑time usage monitoring. According to the company, the API’s transparency and control are designed for production deployments.

Benefits for Enterprises

Multiverse highlights several advantages of its compressed models for business users. Smaller models reduce compute costs, a key consideration as enterprises evaluate alternatives to large language models. They also retain more privacy and resilience because they can operate without constant cloud connectivity, which is valuable for critical fields, drones, satellites and other environments where connectivity is limited. The company notes that its latest compressed model, HyperNova 60B 2602, built on the open‑source GPT‑OSS‑120B, delivers faster responses at lower cost than the original model.

Customer Base and Funding

Multiverse already serves more than 100 global customers, including the Bank of Canada, Bosch and Iberdrola. The firm raised a $215 million Series B last year and is rumored to be preparing a new funding round valued at over €1.5 billion.

Market Outlook

While the CompactifAI app has modest download numbers, Multiverse’s focus appears to be on enterprise adoption rather than mass consumer use. The company’s approach aligns with a broader industry trend toward edge AI, where smaller, efficient models enable new use cases that were previously constrained by cloud dependence.

Used: News Factory APP - news discovery and automation - ChatGPT for Business