OP Pohjola and QuTwo are expanding their Quantum AI Research Center to focus on quantum-inspired AI model compression, a method recently shown to compress seven model families by 90% or more while retaining 91-95% of their quality. The collaboration aims to make open language models more efficient and cost-effective for widespread business use, addressing the increasing expenses of scaling AI adoption.
“Open models and model compression open up new ways for us to adopt AI across our business,” says Timo Ritakallio, President and Group CEO of OP Pohjola, noting the company’s systematic investment in AI since 2017 and quantum research since 2019.
Quantum-Inspired Compression Enables Open AI Model Scalability
QuTwo’s compression method, combining standard techniques with a precision-recovery layer, achieves over 90% compression across seven model families while retaining 91-95% of original quality. This performance, recently demonstrated by the company, allows for substantial reductions in computational demands without significant loss of model accuracy, OP Pohjola says. The approach differs from traditional compression methods by focusing on recovering precision lost during the reduction process, yielding a more accurate compressed model.
OP Pohjola’s sustained investment in both artificial intelligence, beginning in 2017, and quantum research since 2019 culminated in the establishment of the Quantum AI Research Center (QARC) with QuTwo. This long-term commitment reflects a strategic vision to integrate these technologies for practical applications within the financial sector.
The center’s current focus on open language models is driven by a desire to gain greater control over token usage and associated costs, a critical consideration for large-scale AI deployments. “The value of AI does not come from who runs the largest model, but from who puts capable AI to work across their operations,” says Peter Sarlin, Founder and Executive Chairman of QuTwo. The adoption of quantum-inspired compression techniques represents a shift towards more efficient and cost-effective AI solutions.
Unlike methods requiring quantum computers, these techniques operate on existing hardware, offering an immediate pathway to improved performance. This allows OP Pohjola to explore broader applications of AI throughout its business, optimizing operations and enhancing customer experiences.
The company views this as a practical demonstration of quantum AI’s potential, moving beyond theoretical research to deliver tangible benefits, according to the company. “Quantum-inspired compression makes that feasible, and it’s one of the first places where quantum and AI meet in practice,” Sarlin added.
The value of AI does not come from who runs the largest model, but from who puts capable AI to work across their operations.
Peter Sarlin, Founder and Executive Chairman of QuTwo
QARC Advances AI Efficiency for OP Pohjola’s Financial Services
OP Pohjola is using quantum-inspired methods to improve the efficiency of artificial intelligence models, a move enabled by expanded work at its Quantum AI Research Center, QARC. OP Pohjola, Finland’s largest financial services group with over two million owner-customers, anticipates that these advancements will enable more effective and cost-efficient AI deployment across its business.
Source: https://www.qutwo.com/news/op-pohjola-adopts-qutwos-quantum-inspired-ai-compression




