Multiverse Computing Launches Free 60B Compressed AI Model

Multiverse Computing is dealing with one of the biggest problems in artificial intelligence today. This is the high cost and heavy infrastructure requirements of large language models. These models are powerful, but they are often very big and expensive to use, which makes it hard for many companies to use advanced AI systems. Two big problems that stop companies from using AI are that it uses a lot of memory and it costs a lot to set up.

To solve this problem, Multiverse Computing, a Spanish AI company, has launched a free, smaller version of its 60B AI model. This smaller and more efficient model is designed to perform well without using a lot of computing power. By reducing the memory needed and the costs of using it, the company wants to make advanced AI systems easier and more affordable for businesses.

Quick Overview

• Multiverse Computing has launched HyperNova 60B, a free compressed AI model designed to deliver strong performance with significantly lower memory and infrastructure costs.
• Built using its CompactifAI compression technology, the model is about 32GB in size, offers lower latency, and is optimized for efficient enterprise deployment.
• The upgraded HyperNova 60B 2602 version adds better tool-calling, improved agentic coding capabilities, and reduced inference costs for businesses.
• The company positions itself as a European sovereign AI provider, competing with players like Mistral AI while serving enterprise clients such as Bosch and Bank of Canada.
• By offering a free compressed 60B model, Multiverse aims to make advanced AI more accessible, practical, and cost-effective for global enterprises.

What Is Multiverse Computing Doing?

Multiverse Computing makes large AI models smaller without making them too inaccurate. The aim is to make powerful AI systems more practical and affordable for businesses.

CompactifAI Technology

The company developed a compression method called CompactifAI. This technology is based on quantum computing principles and helps to make large models smaller while keeping them effective.

Multiverse has used this approach to make models released by OpenAI smaller and easier to use.

HyperNova 60B Model

The company's latest release is called HyperNova 60B. Developers can now get it for free on Hugging Face.

Key Features

• It is around 32GB in size, which is about half the size of the original base model.
• It uses less memory
• Responds more quickly (lower latency)
• Runs almost as well as the larger Frontier models
• Developers can use this for free

This means that companies can use advanced AI systems without needing very expensive hardware.

Updated Version: HyperNova 60B 2602

Multiverse has also released an improved version called HyperNova 60B 2602.

This updated version offers:

• Better support for tool calls
• The ability to code better
• Lower costs for using AI and automation tools

These improvements are important for businesses that use AI agents or automation tools, where cost and efficiency matter a lot.

Competition in the AI Market

Multiverse says that HyperNova 60B is a strong competitor to models from Mistral AI, including Mistral Large 3.

Both companies are European AI companies that are growing and expanding into new markets. While there is some rivalry between the two companies, they are also similar in that they both focus on big business customers and AI solutions for governments.

Enterprise Customers and Global Presence

Multiverse has grown and now has offices in the United States, Canada and several European countries.

The company works with big businesses like:

• Iberdrola
• Bosch
• Bank of Canada

This shows that its compressed models are already being used in the real world.

Funding and Growth

Multiverse is reportedly talking to investors about raising around €500 million, which would value the company at more than €1.5 billion. The company has confirmed that it is talking to investors, but it has not yet announced the final numbers.

There are also reports that the money it makes each year could be as high as €100 million, but this hasn't been confirmed.

To help you compare:

• OpenAI is said to make around $20 billion every year.
• Mistral AI has reportedly made more than $400 million from its regular customers every year.

Although Multiverse is smaller than these companies, it is growing steadily.

Focus on Sovereign AI

Multiverse says it is a provider of AI solutions for governments. This means offering AI systems that can be used on their own, reducing reliance on foreign technology providers.

The company recently teamed up with the regional government of Aragón in Spain and has received support from Spanish technology development agencies. Since it was first established, it has also benefited from support in the Basque region.

Why This Launch Matters

It is important to release a free compressed 60B model because:

• It makes it cheaper to run large AI models.
• It makes advanced AI easier for companies to use.
• It makes things more efficient without making them perform any worse.
• It helps to grow the European AI infrastructure

As more and more people and companies around the world start using AI, they are looking for solutions that are both powerful and practical. Multiverse Computing is focusing on finding the right balance between performance and efficiency.

Final Thoughts

Multiverse Computing is dealing with one of the biggest problems in AI: how big and expensive models are. By offering a free version of the 60B model, the company is helping businesses use advanced AI without having to spend a lot of money on infrastructure.

If compression technology keeps getting better, it could change how businesses use large language models in the future.

(Source: TechCrunch)

FAQs

1. What is Multiverse Computing?

Multiverse Computing is a Spanish AI startup that develops compressed large language models to make advanced AI systems more efficient and affordable for businesses.

2. What is HyperNova 60B?

HyperNova 60B is a compressed 60-billion-parameter AI model launched by Multiverse Computing. It is designed to deliver strong performance while using less memory and lower infrastructure costs.

3. How is HyperNova 60B different from other large AI models?

HyperNova 60B is smaller in size (around 32GB) compared to many frontier models. It offers lower latency, reduced memory usage, and improved cost efficiency, making it easier for enterprises to deploy.

4. Where can developers access Multiverse Computing’s model?

Developers can access HyperNova 60B for free on Hugging Face, a popular platform for AI model sharing and development.

5. Why is model compression important in AI?

Model compression reduces the size and hardware requirements of large AI systems. This lowers operational costs, improves speed, and makes enterprise AI adoption more practical and scalable.