Multiverse Computing launched a public API with compressed OpenAI, Meta and DeepSeek models

Multiverse Computing, a Spanish startup in quantum computing and AI optimization, has announced the launch of a public API and demonstration application for its compressed language models. The catalog already includes compressed versions of flagship products from four major laboratories at once: OpenAI, Meta, DeepSeek, and Mistral AI. This is the company's first step toward the mass market after several years of predominantly corporate sales.

Multiverse Computing was founded in 2019 in San Sebastián, Spain. The company initially specialized in applying quantum algorithms to optimization problems in the financial sector — insurance calculations, risk management, and portfolio optimization. Over time, it became clear that mathematical methods from the quantum computing toolkit were also applicable to neural network optimization. This very observation led to the creation of a new product direction — compressing large language models while preserving the maximum portion of their capabilities.

Technically, language model compression is not a single method, but an entire class of approaches. Quantization converts neural network weights from float32 to int8 or int4, reducing memory footprint 2-4 times with minimal accuracy loss on most tasks. Pruning removes the least significant neurons and entire network layers, decreasing computational load. Knowledge distillation transfers the behavior of a large model into a smaller, specially trained architecture. In combination, these techniques enable model reduction by 4-8 times while preserving 90-95% quality on typical tasks.

For business, this opens fundamentally new opportunities: running full-fledged LLM inference on devices without expensive GPU clusters — on industrial controllers, medical equipment, corporate servers, and mobile devices. In regulated industries — healthcare, finance, government sector — this is critically important: sensitive data remains within the organization's perimeter and is not transmitted to external cloud APIs.

The launch of a public API fundamentally changes Multiverse Computing's business model. Until now, the company operated predominantly through direct corporate sales: lengthy negotiations, pilot projects, individual integrations with specific clients. An open API shifts the product into self-service mode — a developer can register and start working without contact with the sales department. The demonstration application serves as a product showroom: potential customers can evaluate the quality of compressed models on real tasks before deciding whether to purchase or integrate.

Special attention deserves the inclusion of DeepSeek in the catalog — a Chinese open-source model that became one of the major technological events of early 2026. Interest in it is enormous, but many Western companies and government structures fear dependence on Chinese infrastructure when using the official API. Local deployment of a compressed version through Multiverse Computing solves this problem: the model runs entirely within the corporate perimeter.

The presence of products from four different providers in the catalog also signals the company's ambitions: positioning not as yet another AI laboratory, but as an infrastructure provider agnostic to the entire large model ecosystem. Multiverse Computing is betting on a simple idea: cloud AI services address obvious mass demand, but leave a huge untapped niche — companies and entire industries that, for regulatory, technical, or economic reasons, cannot or do not want to depend on external APIs. A public catalog of compressed models lowers the barrier to entry for this segment to virtually zero.

If the technology demonstrates sufficient quality in real production scenarios, the company has a real chance of occupying a strategically important position in the infrastructure layer of next-generation AI market.