Multiverse Computing has also recently made waves with the launch of two revolutionary AI models, ChickBrain and FlyBrain. These developments are meant to “supercharge” chat AI advances in Internet of Things (IoT) devices. Founded in 2019, the company is headquartered in Europe where they are rattling the AI status quo. Their small but highly capable models can be run completely offline without requiring an internet connection.
ChickBrain features a whopping 3.2 billion parameters. Its deep reasoning capabilities represent a new milestone in AI. This new model has already surpassed its predecessors in a number of common benchmarks. It is on first place in MMLU-Pro, Math 500, GSM8K and GPQA Diamond. FlyBrain’s even smaller counterpart, on the other hand, is set to provide jaw-dropping performance in an even smaller footprint.
Román Orús, co-founder of multinational firm Multiverse Computing, advances quantum physics and computer science to the frontiers. He pushed the uniqueness of their compression technology. He stated, “We have a compression technology that is not the typical compression technology that the people from computer science or machine learning will do, because we come from quantum physics.” This new technique allows the models to be much smaller while still achieving strong performance scores.
The possible uses for ChickBrain and FlyBrain are extensive. And they can work natively on smartphones and smartwatches. This enables developers to provide users with AI functions without requiring an always-on internet connection. As Orús elaborated, “You can run them on premises, directly on your iPhone or on your Apple Watch.” Multiverse Computing’s models provide unparalleled accessibility. This makes them ideal solutions for verticals that require secure, high-performance AI processing at the edge.
In addition to their new models, Multiverse Computing offers compression technology suitable for other forms of machine learning, including image recognition. This flexibility is enabling developers to incorporate the most powerful new AI capabilities into every type of experience. To its credit, the company has made great strides in that direction. They are offering reduced size versions of other open-source models such as Llama 4 Scout and Mistral Small 3.1. They recently released compressed adaptations of OpenAI’s most recent open models as well.
Multiverse Computing’s innovative solutions have attracted attention from major clients like BASF, Ally, Moody’s, and Bosch, signaling strong demand for their technology. The team at the time raised around $250 million in venture financing. This includes €189 million (approximately $215 million) raised in June, which was exclusively focused on its innovative model compression technology CompactifAI.
The business is in active conversations with other leading device and appliance manufacturers. They want to test out how they can use their AI models within these products. As they continue to expand their client base and technological offerings, Multiverse Computing is poised to play a pivotal role in shaping the future of AI in everyday devices.