With the introduction of its CompactifAI public API Portal, Multiverse Computing is extending its quantum‐inspired AI compression technology to the general public. This allows development teams and companies to have direct, self-service access to compressed AI models from labs such as OpenAI, Meta, DeepSeek, and Mistral AI.
How CompactifAI Provides Edge AI
The company’s technology allows for a substantial reduction of model size without sacrificing accuracy; for example, they showed 83%83% compression with zero accuracy loss for a road anomaly detection system. The capability to produce compressed machine learning (ML) algorithms is showcased through the CompactifAI App’s embedded model “Gilda,” which operates locally and offline on devices with enough Random Access Memory (RAM) and storage; if needed, Gilda will automatically route requests to the remote cloud-based model.
The company states that “With the launch of our new API Portal, developers have access to compressed models with the control and transparency they need to use models in production,” as well as provide real-time monitoring of how the models are being utilized. This also addresses a growing demand from enterprises to have lower compute costs and rely less on third-party cloud infrastructure.
Why It Matters For the AI Industry
VC firm Lux Capital is pointing out that there’s a counterparty risk with AI compute commitments. Co-founder Josh Wolfe sent a public memo to Lux’s board of directors suggesting that compute capacity commitments should be confirmed in writing due to the growing financial instability in the AI supply chain.
The solution: On-device models as an alternative to more traditional paths. In addition, Multiverse Computing claims that their HyperNova 60B 2602 model (compressed from OpenAI’s gpt-oss-120b) is now faster and less expensive than the original, which makes it very useful for agentic coding workflows. The company already has over 100 global customers, including the Bank of Canada, Bosch, and Iberdrola, and is reportedly working to raise a €500 million round at over €1.5 billion valuation.
